Professional Documents
Culture Documents
Fr.Geo Pulikkal,
Chairman,
I would like to take this opportunity to thank the organizing team that did an excellent
job of putting this conference together. I am also indebted to our reviewers who reviewed the
manuscripts, sometimes under extreme time constraints, and selected the best papers that fit
this conference.
We hope you will find the conference productive, informative and enjoyable. We also
look forward to receiving your constructive comments that would help us in our future
planning.
Fr.George Asarikunnel,
Bursar,
Vimal Jyothi is established to transform the professional education scenario to suit for
the needs of the time. Knowledge is the asset of humanity and Vimal Jyothi is positioned as the
platform for the knowledge exchange.
It is apt that we host a National Conference on Signal & Image Processing- SIP 2014, in
our campus, which will showcase the modern trends in the technical arena.
I congratulate the team who took the blood shedding pain to materialize this golden
milestone.
Message from Administrator
Fr.Jinu Vadakkemulanjanal,
I joyfully wish to extend my cordial greetings to the participants and Vimal Jyothi family
members for the National Conference on Signal & Image Processing-SIP 2014. SIP provides an
inter-disciplinary venue for researchers and academicians to address the rich space of signal &
image processing and to exchange their views and ideas. It is my earnest wish that this
conference would bring advancements of theory and practice in the field of signal & image
processing and promote collaborative excellence.
It is my earnest wish that this conference would bring advancement of theory and
practice in the field of signal & image processing and promote collaborative excellence.
The success of the conference depends ultimately on many people who have worked in
planning and organizing, both the technical programme and supporting social arrangements. I
congratulate the VJEC family for their thorough and relentless effort.
Principal,
I wish the SIP become a national meeting ground for researchers, engineers, scientists
and specialists in the various research and development fields of signal & image processing. The
conference offers a platform for global experts to gather and discuss intensively on various
topics, ranging from their minute state of aspects to prospects and possible breakthroughs in
the future.
I would like to express my sincere compliments to all authors for their outstanding
contributions to this event and in particular to the members of the review and evaluation panel
for their competent assessment of the large number of submissions and presentations.
Likewise I would also like to express my appreciation to the organizing committee, for their
careful preparation of the sessions.
Message from Organizing Secretary
Prof.Roshini T V,
Organizing Secretary,
SIP aims to provide a platform for research scholars, scientists, industrial professionals
and students to exchange and share their research accomplishments, achievements and new
ideas on contemporary and emerging trends in signal & image processing.
Let me take this opportunity to express our humble gratitude to all the members of
advisory committee, technical committee and various experts who helped with the review who
helped with the review process by devoting their precious time and energy in assessing the
papers and providing suggestions to improve the contents. We would like to express our
sincere thanks to the technical session chairs for disseminating their knowledge thereby making
the conference of a fruitful experience to be cherished.
On behalf of SIP team, I convey our sincere thanks to the management of VJEC, for
providing great logistics and organizational support. Organizing committee members and
students deserve a special mention for lending their timely and whole hearted support.
With immense pleasure and sense of pride, I take this opportunity to welcome all the
eminent guests, experts, scholars, delegates and participants to SIP. Please join us to make the
conference successful and memorable.
TABLE OF CONTENTS
- Anjali Nambiar
Mismatch Condition
- Anupriya Augustine
- Diljo Thomas
Area Network
images in the brain results in a sensation of depth. Stereograms rig is to be calibrated, i.e., the cameras internal parameters,
illustrate this ability of the human mind. Computer stereo mutual position and orientation should be known, to apply
vision tries to imitate this depth perception. The basic idea rectification. After rectification, for each image point in the
is to use two cameras acquiring the same scene. Similarly left image, we can always find its corresponding image point
to our eyes, the two cameras are separated by a distance. A in the right image by searching the same row. Therefore, the
computer analyzes the two images captured and tries to match rectification procedure reduces the point matching procedure
them.The geometry that deals with the stereo images is called from 2D search to 1D search. Thus, we can improve both the
epi polar geometry. matching speed and the matching accuracy.
The basic idea behind the computational stereo problem is
that a real object has multiple projections depending on the II. C AMERA M ODEL
point of view.By having atleast two images from the similar The camera performs a linear transformations from the 3D
view ,computational stereo approach tries to find the depth by projective space P 3 to the 2D projective space P 2 as shown
some comparison.The points of view should be parallel and in fig3. Camera model is the mathematical description of
not too far away. the physics process from scene imaging to image plane. The
The 3D reconstruction problem consists of three consecutive
projective transformation in the general case can be factorized
stages:
into three simpler transformations which corresponds to three
1) Camera Calibration/Rectification: In order to use the transitions between these four coordinate systems.
cameras for the purpose of 3D depth reconstruction
A point,M = [x y z]T , in the 3D space is represented by
the cameras have to be calibrated first.By performing
its homogeneous coordinates,M = [x y z 1]T . Similarly,
Camera calibration internal and external properties of
its image on the retinal (or image) plane, m = [x y]T , is
the camera can be obtained.Using these parameters we
represented by its homogeneous coordinates,m = [x y 1]T .
can make the images undistorted and rectified such
The main property of the camera model is that the relationship
that corresponding epipolar lines lie along the same
between the world coordinates and the pixel coordinates is
horizontal axes.
linear projective. This relationship remains linear regardless
2) Correspondence:It is the process of finding disparity for
the choice of both coordinates.
each and every point from the rectified images,actually
rectification makes this process simpler by limiting the
difference only in X direction.
3) Reconstruction:The disparity is directly related to the
depth From the disparity map,and from the disparity
depth can be calculated.
In order to reconstruct a scene from the camera pair
measurements the system must be calibrated.It is the process
of relating the ideal model of the camera to the actual physical
device and of determining the position and orientation of Fig. 3. Camera cordinates and world cordinates.
the camera with respect to a world reference system [1].
Camera calibration consists of two parts, internal calibration
Thus,M and m are related by
and external calibration.Internal calibration means finding out
the parameters that relates to the camera itself such as optics,
including the focal length, the image centre, the image scaling sm = P M (1)
factor and the lens distortion coefficients and external calibra-
where s, a scale factor, is called the depth; P is a 3 4
tion finds the parameters relates to the pose and orientation of
matrix, called the perspective projection matrix.
two cameras [2].
Because s is an arbitrary scale which could take any value,
The stereo rectification of an image pair is an important
we will simply denote Equation 1 as:
component in many computer vision applications. The precise
3D reconstruction task requires an accurate dense disparity m
= PM (2)
map, which is obtained by image registration algorithms. This
is a property that can also be achieved by precise alignment The camera is modelled by its perspective projection matrix
of the stereo cameras before image capture. By estimating the (PPM) P , which can be decomposed, using the QR factoriza-
epipolar geometry between two images and performing stereo- tion, into the product
rectification, the search domain for registration algorithms is
reduced and the comparison is simplified, because horizontal P = A[R t] (3)
lines with the same y component in both images are in one to
one correspondence. Stereo-rectification methods simulate ro- where A is called the camera calibration matrix which
tations of the cameras to generate two co-planar image planes maps the normalized image coordinates to the retinal image
that are in addition parallel to the baseline [3]. Here the stereo coordinates. This matrix has a number of entries called the
Natonal Conference on Signal & Image Processing -2014 (SIP) 3
where r is the aspect ratio, f is the focal length, is the restricts the search domain for each match to a line parallel
skew, [u0 , v0 ]T is the principal point, R is the rotation matrix to the x-axis. Due to the redundant degrees of freedom, the
and T is the translation vector. The camera is called non-skew solution to stereo rectification is not unique and actually can
if = 0, aspect-free if r = 1 and with Euclidean image plane lead to undesirable distortions [4] or be stuck in a local
when both = 0 and r = 1. The camera calibration matrices. minimum of the distortion function. Image rectification is
usually performed regardless of camera precision, because
rf 0 u0
f u0
f 0 u0
it may be impractical to perfectly align cameras, and even
0 f v0 0 f v0 and 0 f v0 perfectly aligned cameras may become misaligned over time.
0 0 1 0 0 1 0 0 1 The stereo rectification of an image pair is an important
component in many computer vision applications. The precise
are called non-skew, aspect-free and Euclidean calibration 3D reconstruction task requires an accurate dense disparity
matrices respectively. map, which is obtained by image registration algorithms. By
III. STEREO CAMERA CALIBRATION estimating the epipolar geometry between two images and per-
forming stereo-rectification, the search domain for registration
The camera calibration is carried out using the Camera
algorithms is reduced and the comparison simplified, because
Calibration Toolbox for Matlab . The left and right cameras
horizontal lines with the same y component in both images
are calibrated separately one after another. The results obtained
are in one to one correspondence. Stereo-rectification methods
from the left and right camera calibration is used as the input
simulate rotations of the cameras to generate two coplanar
to calculate the stereo camera calibration.
image planes that are in addition parallel to the baseline.
The idea behind rectification is to define two new PPMs
obtained by rotating the old ones around their optical centers
until focal planes becomes coplanar, thereby containing the
baseline. This ensures that epipoles are at infinity; hence,
epipolar lines are parallel. To have horizontal epipolar lines,
the baseline must be parallel to the new X axis of both
cameras[5]. In addition, to have a proper rectification, con-
jugate points must have the same vertical coordinate. This is
obtained by requiring that the new cameras have the same
intrinsic parameters. Note that, being the focal length the same,
retinal planes are coplanar too.
along the x-axis). [4] Joint depth and color camera calibration with distortion correction Daniel
2) Apply the same rotation to the right camera to recover Herrera C., Juho Kannala, and Janne Heikkila.
[5] Analysis of Stereo Image Rectification Algorithms Zhiwei Zhu, Ming
the original geometry. Jiang, Chih-ting Wu.
3) Rotate the right camera by R. [6] Huihuang Su, Bingwei He, Stereo rectification of calibrated image pairs
4) Adjust the scale in both camera reference frames. based on geometric transformation, I.J.Modern Education and Computer
Science, 2011, 4, 17-24
Reconstruction of 3D points by triangulation be performed
from the rectified images directly, using the two new PPMs.
Figure 5 are the original image before rectification and after
rectification. The horizontal lines in rectified images indicates
the left and right image is horizontally aligned compared to
the original images.
V. R ESULTS
The hardware architecture includes the two camera system
which are similar. The installation of the two camera system
is in parallel while the baselines length of two cameras is
kept 7.5cm. A pair of USB web cameras are used as stereo
vision module. The stereo camera pair, collects synchronized
pairs of images, labelled left and right. Both pairs of images
are calibrated using the Camera Calibration Toolbox.
In Fig 6 the stereo left image and right image captured are
shown. The images after rectification is also shown in the same
figure.After rectification it could be seen that in both images
points images corresponding to same object will align in same
X axis .
VI. C ONCLUSION
Image rectification is an important task in machine vi-
sion,whose goal is to make the image points corresponding
to same real point aligned in the straight line so that disparity
calculation becomes simpler.The main advantage is that we
need to search in only one direction during the disparity
calculation.
R EFERENCES
[1] Zhang, Z., A flexible new technique for camera calibration. IEEE trans-
actions on Pattern Analysis and Machine Intelligence, 2000.
[2] Liangfu Li, Zuren Feng, Yuanjing Feng, Accurate Calibration of Stereo
Cameras for Machine Vision. JCS &T Vol. 4 No. 3, JCS&T Vol. 4 No.
3
[3] A Fusiello, E Trucco, A Verri, A compact algorithm for rectification of
stereo pairs, Machine Vision and Applications, vol. 12, pp.1622, July
2000.
Natonal Conference on Signal & Image Processing -2014 (SIP) 6
AbstractUltra-wideband (UWB) is a new promising technol- within 3.1 to 10.6 GHz band. FCC has also defined the power
ogy that has the capability to revolutionize the field of wireless spectral density (PSD) of UWB to be less than 41.3 dBm/MHz
communications. These narrow pulses spread the energy over within 3.1 to 10.6 GHz, the same as FCC part 15 limit. Al-
a wide frequency band, and hence the name ultra-wideband
(Impulse radio). As one of the major concern while choosing though Ultra Wide Band (UWB) wireless systems supposed to
a pulse shape for UWB communication is that it should meet coexist with other spectrally overlapping (narrowband) wire-
the FCC emission mask requirements. Ultra Wide Band (UWB) less systems, mutual interference degrades the performance of
wireless systems supposed to coexist with other spec- trally both system [1] . Specifically, IEEE 802.11a wireless LAN
overlapping (narrowband) wireless systems, mutual interference users experience performance degradation due to increment in
degrades the per- formance of both systems. In this work, we
propose two techniques for reducing this mutual interference the noise floor due to the presence of UWB systems.TR-UWB
which do not require modification of the narrowband system. We systems are especially susceptible to narrowband interference
pro- pose to shape the spectrum of the Transmitted Reference because the interference will be multiplied at the receiver
(TR)-UWB signal to avoid its interference to other wireless end. The proposed method improves the UWB system to
systems using wavelet based multi-resolution technique . The combat the interference from and to the UWB signals by
reverse (narrowband) interference is alleviated using a notch
filter at UWB receiver that could have much smaller bandwidth appropriate signal processing at the UWB side only. To reduce
in this Gaussian pulse-shaped multi-resolution environment.In the interference level two different strategies are used in this
this method generation of wavelet based impulse waveform is work. The most widely studied and implemented pulse shape
implemented in MATLAB for obtaining an interference free for single band UWB communications is the first derivative
multisystem. of a Gaussian pulse, called a Gaussian monocycle. One of the
Index TermsWireless communication, UWB, WLAN, FIR,
IIR, Wavelet transform, Spectral overlap, Multi-resolution, Pulse
original leaders in UWB research and development is Time
shaping, Coexistence. Domain Corporation (TDC). Since about 2002 TDC has used
the Gaussian monocycle in binary phase shift keying (BPSK),
I. I NTRODUCTION or bi-phase modulation, and quadrature flip time modulation
(QFTM) in the design, implementation and test verification of
Ultra Wide Band (UWB) is emerging as the most promising its PulsON UWB radio.
short range high-speed wireless solution because it has the
1) Suppress interference by not using the frequency band
potential to transmit up to 300 Mbps over short distances.
where the interferer operates. This is done by performing
UWB is expected to have large application in office build-
precise pulse shaping to the UWB signal using a multi-
ings, conference rooms, and public places (airports, shopping
resolution wavelet technique.
malls, etc.). The key advantage of UWB is the mitigation
2) Cancel narrowband interference to the UWB system.
of spectral congestion because UWB systems are designed
This is done by introducing a notch filter at the UWB
to coexist with other spectrally overlapping wireless systems.
receiver.
UWB also consumes very little power, saving battery and is
inherently secure due to noise-like wideband transmission. Wavelet literaly mean a small wave.A transform is a
In addition, UWB radio can penetrate obstacles providing mathematical operation that takes a function or sequence
detailed radarview from the wall helping to locate the hidden and maps it into another one.The basis functions of the
objects with high precision. Therefore, UWB has applications wavelet transform (WT) are small waves located in different
in ground penetrating radar, medical imaging, construction, times.Wavelet tranforms are obtained using scaling and trans-
mining and surveillance systems. UWB is also useful in ve- lation of a scaling function and wavelet function. Therefore,
hicular radar systems that can be integrated into the vehicular the wavelet transform is localized in both time and frequency.
navigation system. Examples are collision warning radars, The wavelet transform provides a multiresolution system by
improved airbag activation, and field disturbance sensors. using a variable length window. Multiresolution is useful in
An Ultra Wide Band signal is currently defined by FCC as several applications.
a signal that occupies more than 500 MHz bandwidth and lies Multi Resolution Analysis refered to analyzing a signal
Natonal Conference on Signal & Image Processing -2014 (SIP) 7
the pulse width reduces both: the overlapping with other pulses 1) Interference reduction from UWB to 802.11a: Impulse
and interferences with other systems. radio UWB signals are often studied using second and third
FIR based pulse shaping of UWB radio signal is obtained by derivatives of Gaussian function. A short Gaussian waveforms
FIR filter coefficients by employing butterworth approximation is used to obtain large bandwidth. A Gaussian waveform, in
method. The error between the desired frequency response and general,can be described by,
the output filter frequency response in the desired bandwidth
t2
have been minimized with minimax criterion, which minimize p(t) = Aexp( )) ; t 0 (1)
2
the maximum error over the frequency band of interest by
using the Parks-McClellan algorithm. The Parks- McClellan To increase the flexibility and to reduce the sensitivity to
(PM) algorithm is based on the Alternation Theorem. It is interference is splitting the frequency spectrum into sub-bands.
utilized in polynomial approximation theory and used as the With this approach, the UWB signal is divided into a number
traditional FIR design algorithm. It can also be used to achieve of pulses. Each pulse has a different center frequency but
linear-phase filters and an equi-ripple design. Even though a the same pulse width. The total UWB signal is the sum of
good approximation of the FCC mask is obtained using the all pulses.Wavelets can be used to perform a multi-resolution
PM algorithm, it is not an optimal solution. This is due to analysis of a transmitted UWB waveform into multiple sub-
the fact that the filter length and, accordingly, the pulse length band pulses centered on set of different subcarriers data can be
becomes very large and can cause overlapping of the pulses transmitted through all the sub-bands or through only selected
and increase energy consumption and interferences. Hence, in number of them.
order to achieve an optimal solution, we must reduce the filter t2
length and at the same time obtain an accurate approximation f (t) = Aexp( )sin(2fc t)) (2)
2
of the FCC mask. And hence there is several limitations for
FIR and IIR based pulse shaping of UWB radio signal such as t t2
the filter length will affect the optimal solution, pulse length ( ) = exp( 2 )sin(2fc t)) (3)
a
will be high, energy consumption is high,and also occurance
of noisy template problem and hence additional enhancement
techniques are needed.
B. Wavelet-based spectral shaping of UWB radio signal for
multisystem coexistence
Ultra Wide Band (UWB) wireless systems are supposed
to coexist with other spectrally overlapping (narrowband)
wireless systems, mutual interference occurs which degrades
the performance of both systems.There is two techniques for
reducing this mutual interference which do not require modifi-
cation of the narrowband system. By shaping the spectrum of
the Transmitted Reference (TR)-UWB signal using wavelet
based multi-resolution technique, its interference to other
wireless can be avoided. The reverse (narrowband) interference
is alleviated using a notch filter at UWB receiver that could
have much smaller bandwidth in the Gaussian pulse-shaped
multi-resolution environment.
Fig. 2. The gaussian pulse
IV. WAVELET BASED PULSE SHAPING FOR UWB
C OMMUNICATION Assume a nfs, with n an integer and pulse repetition rate fs
400 MHz. Then we can approximate the transmission pulse
Since the previous methods like FIR and IIR pulse shaping
with that of 20 subcarriers from 3.2 to 10.2 GHz. The designed
is having several drawbacks, the wavelet based pulse shaping is
cumulative pulse and its power spectral density (PSD) are
a promising technique. Wavelet based multi resolution spectral
shown in in Fig. 2 and Fig. 3.
shaping technique is to be designed for shaping TR UWB
signal and in the UWB receiver side a notch filter need to be X t
designed. f (t) = Wf (a)( ) (4)
a
A. Interference reduction from UWB to 802.11a Where f(t) is the pulse waveform and ( at ) represents
Impulse radio UWB signals are often studied using second subcarriers. The total pulse waveform is the weighted sum
and third derivatives of Gaussian function. A short Gaussian of sub-band pulses.
waveforms with typically less than 1 ns pulse width is used 27
to obtain large bandwidth. A Gaussian waveform, in general,
X t
S(t) = Wf (nfs )( (5)
can be described by, 7
nfs
Natonal Conference on Signal & Image Processing -2014 (SIP) 9
Fig. 3. PSD of the gaussian pulsee Fig. 5. PSD of the multi-carrier pulse
Fig. 4. multi-carrier pulse PSD simulation are carried out for the 5th, 6th and 7th
derivative of the Gaussian pulse for optimum values and
for values greater than and less than optimum value are also
To reduce interference from the UWB system to IEEE calculated. Fig. 7 illustrates the simulation result for the 7th
802.11a WLAN systems, we eliminated five interfering sub- derivative of the Gaussian pulse for various values of and
carriers around 5.25 GHz. This reduces transmission signal in shows that the PSD of the pulses for the optimal values of the
IEEE 802.11a band. The resultant pulse and its corresponding pulse shape factor satisfies the FCCs mask closely as compared
PSD are shown in Fig. 4 and Fig. 5. In the case of coexistence to other taken nearby values of .
with IEEE 802.11a, a number of UWB sub-bands centered Also we have found that for n=7 the optimum value of
around the interferer (5 GHz) are removed to guarantee no does not closely matches the FCC mask at one of the corner
overlapping between 802.11a and UWB signals. frequency at f=10.6GHz. For this reason we limited the
2) Interference reduction from IEEE802.11a to UWB: waveforms to the 6th order since higher order derivatives are
Interference suppression in the TR receiver is achieved by not practical. In fact, even theoretically the Gaussian pulse
eliminating the interfering band by a notch filter. Gaussian can be infinitely differentiated; it is usually limited to 5 or
pulse has the advantage again because notch filter bandwidth maximum 6 derivatives.
can be relatively narrow compared to the full-band of Gaussian
pulse. Notch filter effectively causes the interfering band to be V. R ESULT AND D ISCUSSION
turned off.It reduces interference to the UWB system from The system performance with four different IEEE channel
the IEEE 802.11a WLAN system. UWB receiver module models with and without NBI. The performance of the original
retrieves the baseband data from the received signal. At the TR receiver was very bad due to both NBI and multipath con-
demodulation, the baseband signal is received and correlated ditions. However,the proposed architecture is able to improve
Natonal Conference on Signal & Image Processing -2014 (SIP) 10
TABLE I
O PTIMAL VALUES OF P ULSE SHAPE FACTORS FOR G AUSSIAN technique. And also wavlet based shaping of UWB pulse is
DERIVATIVES having several advantages like energy efficiency, less com-
plexity, high data capacity, low power, resilient to distortions
Gaussian Pulse Shape Peak Frequency PSD(dm/MHz)
Derivative Factor and fading.Another important advantage again is notch filter
fifth 0.177e-9 7.139 -42.65 bandwidth can be relatively narrow compared to the full-
sixth 0.188e-9 7.389 -42.98 band of Gaussian pulse and thus wavelet based UWB based
seventh 0.1988e-9 7.515 -41.76
pulse shaping is the promising and most efficient In proposed
method both interference in TR- Transmitter and TR- Re-
ceiver is removed by wavelet-based spectral shaping for UWB
impulse radio and the reverse (narrowband) interference is
alleviated using a notch filter at UWB receiver. This is to be
implemented in MATLAB or SIMULINK.
R EFERENCES
[1] K. Ohno, T. Ikebe, and T. Ikegami, A proposal for an interference
mitigation technique facilitating the coexistence of biphase uwb and
other wideband systems, in Ultra Wideband Systems, 2004. Joint with
Conference on Ultrawideband Systems and Technologies. Joint UWBST
IWUWBS. 2004 International Workshop on, 2004, pp. 5054.
[2] Z. Tian and G. Giannakis, A wavelet approach to wideband spectrum
Fig. 7. PSD for 7th derivative of Gaussian Pulse for different values sensing for cognitive radios, in Cognitive Radio Oriented Wireless
Networks and Communications, 2006. 1st International Conference on,
2006, pp. 15.
[3] X. Zhou, H. Zhang, and I. Chlamtac, Transmit power allocation among
the performance when the multipath is reasonably low (in the pswf-based pulse wavelets in cognitive uwb radio, in Cognitive Radio
Oriented Wireless Networks and Communications, 2006. 1st International
case of CH1 and CM2). Thus it is possible to conclude that Conference on, 2006, pp. 15.
our proposed wavelet spectral shaping plus the notch filter [4] K. Ohno, T. Ikebe, and T. Ikegami, A proposal for an interference
scheme significantly alleviates the interference from WLAN mitigation technique facilitating the coexistence of biphase uwb and
other wideband systems, in Ultra Wideband Systems, 2004. Joint with
to UWB system when multipath is insignificant.Thus a good Conference on Ultrawideband Systems and Technologies. Joint UWBST
channel impairment mitigation technique is needed to enjoy IWUWBS. 2004 International Workshop on, 2004, pp. 5054.
the full benefit under multipath conditions. It is important to [5] R. Hoctor and H. Tomlinson, Delay-hopped transmitted-reference rf
communications, in Ultra Wideband Systems and Technologies, 2002.
observe that the interference from the UWB to WLAN system Digest of Papers. 2002 IEEE Conference on, 2002, pp. 265269.
is also negligible with this approach, because essentially there [6] J. Choi and W. Stark, Performance of ultra-wideband communications
is no emission in the overlapping band. Furthermore, note with suboptimal receivers in multipath channels, Selected Areas in
Communications, IEEE Journal on, vol. 20, no. 9, pp. 17541766, 2002.
that just the notch filter cannot be applied (at the UWB [7] Y.-L. Chao and R. Scholtz, Optimal and suboptimal receivers for ultra-
receiver) without the wavelet pulse shaping at the transmitter wideband transmitted reference systems, in Global Telecommunications
because it would reject a useful portion of the UWB signal too. Conference, 2003. GLOBECOM 03. IEEE, vol. 2, 2003, pp. 759763
Vol.2.
This technique does not reduce the throughput of the UWB [8] S. Shiwei Zhao, H. Liu, and Z. Tian, A decision-feedback autocorrelation
system. It can be used to alleviate any narrowband interference receiver for pulsed ultra-wideband systems, in Radio and Wireless
to UWB, not just IEEE 802.11a. However, it introduces Conference, 2004 IEEE, 2004, pp. 251254.
[9] F. Dowla, F. Nekoogar, and A. Spiridon, Interference mitigation in
additional signal processing complexity at the transmitter and transmitted-reference ultra-wideband (uwb) receivers, in Antennas and
receiver sides. This complexity need to be quantified in the Propagation Society International Symposium, 2004. IEEE, vol. 2, 2004,
future. The biggest challenge faced in the work was containing pp. 13071310 Vol.2.
the pulse width within 2 ns in time domain while performing
the required spectral shaping.
VI. A PPLICATIONS
UWB has applications in ground penetrating radar, med-
ical imaging, construction, mining and surveillance systems.
UWB is also useful in vehicular radar systems that can be
integrated into the vehicular navigation system. Examples are
collision warning radars, improved airbag activation, and field
disturbance sensors.
VII. C ONCLUSION
In a multisystem communication mutual interference de-
grades the performance of both systems. Since the previous
methods like FIR and IIR pulse shaping is having several
drawbacks, the wavelet based pulse shaping is a promising
Natonal Conference on Signal & Image Processing -2014 (SIP) 11
Abstract:-This paper aims to investigate the performance of etc. The international standard organization (ISO) has proposed
wavelet tree based image compression system. In wavelet the JPEG standard [1] for image compression. These standards
packet tree based compression system, compression is done by employ discrete cosines transform (DCT) to reduce the spatial
using various entropies. In this paper an image is compressed redundancy presented in images. It is noted that DCT has
using these different entropies and their results are compared to drawbacks of the blocking artifacts, mosquito noise and aliasing
select the best entropy for the given image. Also performance of distortions at high compression ratio [2].
wavelet packet tree based image compression system and Discrete Wavelet Transform (DWT) has emerged as a
discrete wavelet transform based compression system are popular technique for image coding applications. DWT has
compared. The test results indicate that wavelet packet tree high decorrelation and energy compaction efficiency. The
based image compression system shows good performance as blocking artifacts and mosquito noise are absent in a wavelet-
compared to discrete wavelet transform compression system. based coder due to the overlapping basis functions. The
JPEG 2000 standard employs a discrete wavelet transform
Keywords: Wavelet, Wavelet Packet Tree, Shannon, for image compression due to its merits in terms of scalability,
Threshold, Norm, Log energy Entropy localization and energy concentration [3].
The performance of discrete wavelet transform based coding
1. Introduction depends on the wavelet decomposition structure. In wavelet
decomposition, the approximate component of image is further
Visual communication is becoming increasingly important with
decomposed, but in wavelet the approximation as well as
detailed components are decomposed. In this paper we will
applications in several areas such as multimedia, communication,
investigate the performance of wavelet packet tree based image
transmission and storage of remote sensing images, education and
compression system by using different entropy-based criterion.
business documents, and medical images
2. Image compression using discrete basis functions are obtained from a single photo type
wavelet transform wavelet called the mother wavelet by dilation (scaling) and
Compression is achieved by reducing or removing redundancy.
translation (shifts). The discrete wavelet transform for two
The redundancy of an image is caused by the correlation dimensional signals can be defined as follows.
between pixels, so transforming the pixels to a representation
where they are decorrelated eliminates the redundancy. It is also
W (a1,a2,b1,b2,) = 1/a1a2 * X-b1/a1 , X-b2/a2
possible to think of a transform in terms of the entropy of the
image. In a highly correlated image, the pixels tend to have
equiprobable values, which results in maximum entropy. If the
The indexes w(a1,a2,b1,b2,) are called wavelet coefficients of
transformed pixels are decorrelated, certain pixel values become
signal X, and a1,a2 are dilation and b1,b2 is translation , is
common, thereby having large probabilities, while others are
the transforming function, which is known as mother wavelet.
rare. This results in small entropy. Quantizing the transformed
Low frequencies are examined with low temporal resolution
values can produce efficient lossy image compression. Wavelet
while high frequencies with more temporal resolution. A
transforms have received significant attention because their multi
wavelet transform combines both low pass and high pass
resolution decomposition allows efficient image analysis.
filtering in spectral decomposition of signals.
Wavelets are functions that
In case of discrete wavelet, the image is decomposed into a
satisfy certain mathematical requirements and are used in
discrete set of wavelet coefficients using an orthogonal set of
representing data or other functions. The basic idea of the
basis functions. These sets are divided into four parts such as
wavelet transform is to represent any arbitrary signal X as a
approximation, horizontal details, vertical details and diagonal
superposition of a set of such wavelets or basis functions. These
Natonal Conference on Signal & Image Processing -2014 (SIP) 12
details. Further decomposition of approximation takes The wavelet decomposition tree is a part of this complete
place ; we get again four components shown in Fig.1. binary tree. For instance, wavelet packet analysis allows the
signal S to be represented as A1 + AAD3 + DAD3 + DD2.
This is an example of a representation that is not possible
LL2 HL2 HL1 with ordinary wavelet analysis. Choosing one out of all these
possible encodings uses an entropy-based criterion to select
LH2 HH2 the most suitable decomposition of a given image.
LH1 HH1
A3 D3 4. Experimental Results
.
Fig. 2. wavelet decomposition tree In this paper results of compression are compared in
terms of percentage of zeros, percentage of energy
S A1 + D1 = A2 + D1 + D2 retained and Root Mean Square Error. Root Mean Square
Error (RMSE) is square root of the cumulative squared
= A3 + D1 + D2 + D3 error between the compressed and the original image. The
mathematical formulae for the RMSE is:
In wavelet packet analysis, the details as well as the
approximations can be split. This yields more than Error E = Original image Reconstructed image ..i
different ways to encode the image. This is the wavelet
packet decomposition tree. RMSE = sqrt[E / (SIZE OF IMAGE)].ii
S A lower value for RMSE means lesser error. The energy retained
describes the amount of image detail that has been kept; it is a
measure of the quality of the image after compression. The
A1 D1 number of zeros is a measure of compression. A greater
percentage of zeros implies that higher compression rates can
AA2 DA2 be obtained. So, a compression scheme having a lower RMSE,
AD2 DD2
higher percentage of zeros and higher percentage of energy
Fig. 3.wavelet packet decomposition tree retained can be recognized as a betterone.
Natonal
To select the best entropy Conference
in wavelet packet on
treeSignal
based & Image Processing -2014 (SIP) 13
image compression system, we performed the compression
on image wbarb using different entropies. Fig. 4 (a) shows Type of Percentage Percentage RMSE
the original wbarb and Fig. 4(b) ,4(c), 4(d), 4(e) shows results entropy of of energy
of compressed image using Shannon, threshold, norm, log zeros retained
energy and sure entropy. Table1 shows the results of using
various types of entropy for image wbarb. shannon 86.17 99.52 6.726
Then to compare the wavelet packet tree and discrete wavelet
transform based compression systems we performed.the threshold 85.70 99.54 6.541
compression on standard 256*256 color images: leena, woman,
wbarb, trees and julia constructed images using the wavelet norm 86.12 99.54 6.564
packet tree and discrete wavelet transform based compression
systems. Results are observed in terms of percentage of zeros, Log 85.74 99.55 6.515
percentage of energy retained and Root Mean Square Error. Fig. energy
5(a) shows the original image and Fig.5 (b), Fig.5(c) shows the
compressed image using wavelet decomposition and wavelet Table1: Results of using different types of entropies in
packet best tree respectively for the image woman. Table2 wavelet packet best tree for image wbarb.
shows the results of discrete wavelet transform and Table3
shows the result of wavelet packet best tree for different images. Name of Global Percentage Percentage RMSE
the threshold of of energy
Original image images value zeros retained
6. REFERENCES
AbstractCognitive radio (CR) is a novel concept that allows the other CBD algorithms. The initial step of the PC algorithm
wireless systems to sense the environment, adapt, and learn from is similar (to some extent) to the other CBD algorithms i.e.
previous experience to improve the communication quality. In calculation of sample covariance matrix of OFDM signals.
this paper, orthogonal frequency division multiplexing (OFDM)
technique is investigated as a candidate transmission technology II. S YSTEM M ODEL
for CR. Since orthogonal frequency-division multiplexing
(OFDM) is one of the major wideband transmission techniques, We consider a multiple receive antenna system, where each
spectrum sensing for OFDM based primary signals using its antenna is connected to an independent Radio Frequency (RF)
sample covariance matrix is developed. This paper presents a front-end and the inter-antenna distance is less than half of
novel technique for blind signal detection that uses Principal the wavelength of the Licensed Users (LU) centre frequency
Component (PC) Analysis.
being received. This enables correlated signal reception at each
Keywords-Cognitive Radio, Orthogonal Frequency Division antenna. The signal is sampled atfs W where fs is the
Multiplexing, Principal Component Analysis, Covariance Matrix, sampling frequency and W is the bandwidth of the received
Spectrum Sensing signal. Let Ts = 1/fs be the sampling period. We define
zi (n) = zi (nTs ), si (n)
= si (nTs ) and wi (n) = wi (nTs )
I. I NTRODUCTION where zi (n) is the received signal at the ith receive antenna,
With emerging technologies and with the increasing number si (n) is the version of LUs signal received at the ith receive
of wireless devices, the radio spectrum is becoming increas- antenna after passing through the channel and wi (n) is the
ingly congested everyday. On the other hand, measurements independent and identically distributed (i.i.d) white noise
show that wide ranges of the spectrum are rarely used most (AWGN) added at the ith front-end. Each antenna is connected
of the time, while other bands are heavily used. Depending with an independent RF front-end and the RF front-ends are
on the location, time of the day, and frequency bands, the synchronized for synchronous reception of the LU signals. The
spectrum is actually found to be underutilized. However, those receiver multiplexes the complex baseband samples received
unused portions of the spectrum are licensed and thus cannot from R front-ends to make a single stream of N complex
be used by systems other than the license owners. Hence, there samples that can be expressed as
is a need for a novel technology that can benefit from these Z = [zi (n), ..., zR (n), zi (n 1), ...zR (n 1),
opportunities. cognitive radio (CR) arises to be a tempting
solution to spectral crowding problem by introducing the ...zi (n N/R + 1), ..., zR (n N/R + 1)] (1)
opportunistic usage of frequency bands that are not heavily
occupied by licensed users (LU).
A special case of multi-carrier transmission known as N samples are used to make a single sensing decision about
orthogonal frequency division multiplexing (OFDM) is one the presence of an LU where each RF front-end contributes
of the most widely used technologies in current wireless N/R samples. The process of signal detection requires distin-
communications systems. OFDM has the potential of fulfilling guishing between the following two hypothesis:
the aforementioned requirements of CR inherently or with H0 : Z(n) = w(n)
minor modifications. Because of its attractive features, OFDM (2)
has been successfully used in numerous wireless standards and H1 : Z(n) = s(n) + w(n)
technologies. Principal Component Analysis is a multivariate whereH0 is the null hypothesis that indicates the absence of
analysis technique which is used to reduce the dimensionality the signal and H1 is the alternative hypothesis that indicates
of a data set consisting of a large number of interrelated the presence of the signal. z(n) is the observed signal belonging
variables, while preserving as much as possible of the variation to z. s(n) is the LUs transmitted signal that passes through
present in the data set. The PCA based algorithm (or PC a wireless channel, undergoing multipath fading and path
algorithm for brevity) is very promising in comparison with loss effects. w(n) is the AWGN noise that follows a normal
Natonal Conference on Signal & Image Processing -2014 (SIP) 16
distribution w(n) N (0, 2 n) where 2 n is the noise We assume that X has a zero mean vector. Let 1 2
variance. 3 .... 2R be the ordered eigenvalues (characteristic roots)
Two probabilities characterize the performance of spectrum of C, such that the following condition is satisfied.
sensing process: 1) probability of detection Pd , which is the
|c I| = 0 (6)
probability of correctly detecting the LUs signal in the case of
H1 , 2) probability of false alarm Pf a , which is the probability where I is the identity matrix having the same dimensions
of falsely detecting an LU signal in the case of H0 . It is desired as C. Let 1 ,2 ,3 ,.....2R be the normalized eigenvectors
to maximize Pd (ideally to 100) while minimizing the Pfa (characteristic vectors) of C Let F be a feature matrix, that
(ideally to 0). A low Pd can lead to interference with the LUs is composed of the k most significant eigenvectors, where 1
transmission if the CR node transmits while assuming that the k 2R, i.e.
LU is not in operation. A high Pf a can cause underutilization F = [1 , 2 , 3 .......k ] (7)
of the vacant frequency band of the LU, resulting in a low
efficiency of the CR system. Most significant k eigenvectors correspond to the k highest
ordered eigenvalues. The new transformation of the original
III. P RINCIPAL COMPONENT ANALYSIS data set into PCs can now be defined as
p i = F T xi , i = 1, 2, ...N (8)
This section presents an innovative technique for blind
signal detection that is based on principal component analysis. These PCs give an orthogonal linear transformation of the
PCA is a multivariate analysis technique which is used to original data set. The complete set of PCs can be expressed
reduce the dimensionality of a data set consisting of a large as
number of interrelated variables, while preserving as much P = [p1 , p2 , p3 , ....pN ] (9)
as possible of the variation present in the data set . This
reduction is attained by transforming to a new set of variables, P has k rows and N columns or equivalently it can be said
the principal components (PCs), which are uncorrelated and that it has k principal components. In P the first raw is the
ordered so that the first few PCs retain most of the variation one with the largest variance and is uncorrelated with all the
present in the original data set. remaining raws of P. Similarly the second raw is the one with
The computation of PCs involves the eigenvalue decom- the second largest variance and is uncorrelated with all other
position of a positivesemidefinite symmetric matrix, which is raws of P, and so on.. We can now present a criteria for
the covariance matrix in our case. Two conditions should be distinguishing between signal and noise based on the above
met before PCA is used. Firstly, the data should have a zero discussion for k PCs as
mean and secondly it should follow a normal distribution. The
data considered here is composed of time domain complex k
baseband samples and each of the I and Q components of the
X
D = 1/N (j 21 + j 22 + .... + j 2N ) H
H1
0
(10)
complex samples follow a normal distribution, both in case j=1
of H0 and H1 . The complex baseband samples from each
antenna source are first decomposed into their constituent I where j i is the ith variable of the jth PC and is the
and Q components and are expressed as detection threshold. is determined empirically according to
the desired Pfa.
X = [x1I , x1Q , x2I , x2Q , .....xRI , xRQ ]T (3)
IV. E XPERIMENTAL S ETUP
where R is the number of receive antennas.Due to syn- Principle component analysis based spectrum sensing is
chronous reception, all the decomposed complex baseband simulated using MATLAB. The OFDM signal used in the
samples in the column vector correspond to the same time simulations will focus in the 2k mode of the DVB-T standard.
instant. Multiplexing received samples from different antenna This particular mode is intended for mobile reception of
sources increases spatial and temporal correlation in the data. standard definition DTV. The transmitted OFDM signal is
N complex samples, used in making a single sensing decision, organized in frames. Each frame has a duration of TF , and
form a 2R dimensional data set that is expressed in the form consists of 68 OFDM symbols. Four frames constitute one
of a matrix as super-frame. Each symbol is constituted by a set of K=1,705
carriers in the 2k mode and transmitted with a duration TS .
X = [x(n), x(n 1), .....x(n N + 1)] (4)
A useful part with duration TU and a guard interval with a
For convenience we will express x(n i) as xi . We define duration Mcompose TS .
a sample covariance matrix of the data matrix X as
N
Awgn noise is added to the above mentioned input signal
X at various SNR values ranging from -20db to 10 db. The
C = 1/N xi .xTi (5)
programme is run for 100 times to get the SNR vs probability
i=0
Natonal Conference on Signal & Image Processing -2014 (SIP) 17
V. C ONCLUSION
CR is an exciting and promising technology that offers
a solution to the spectrum crowding problem. On the other
hand, OFDM technique is used in many wireless systems
and proven as a reliable and effective transmission method.
OFDM can be used for realizing CR concept because of its
inherent capabilities that are discussed in detail in this paper.
By employing OFDM transmission in CR systems; adaptive,
aware and flexible systems that can interoperate with current
technologies can be realized.In this paper, a novel blind signal
detection algorithm based on principal component analysis
has been proposed. Under real conditions, the PC algorithm
believed to be performs better than other algorithms.
R EFERENCES
[1] Weili Pan, Haiquan Wang and Lei Shen Covariance Ma-
trix Based Spectrum Sensing For OFDM Based Cognitive Ra-
dio,2012 International Conference on Systems and Informatics
(ICSAI 2012),
[2] Farrukh Aziz Bhatti, Gerard B. Rowe and Kevin W. Sowerby,
Spectrum Sensing using Principal Component Analysis, 2012
IEEE Wireless Communications and Networking Conference:
PHY and Fundamentals
[3] Mitola, J. and J. Maguire, G. Q., Cognitive radio: making
software radios more personal IEEE Personal Commun. Mag.,
vol. 6, no. 4, pp. 1318, Aug. 1999.
[4] Haykin, Cognitive radio: brain-empowered wireless communi-
cations, IEEE J. Select. Areas Commun .,vol.23,no.2, pp.201-
220,Feb. 2005
[5] T. Yucek and H. Arslan, A survey of spectrum sensing algo-
rithms for cognitive radio applications, ,IEEE Commun. Surveys
and Tutorialsvol. 11, no. 1, pp. 116130, 2009.
[6] Y. H. Zeng and Y.C. Liang, Eigenvalue based spectrum sensing
algorithms for cognitive radio, IEEE Trans. Commun., vol. 57,
no. 6, pp. 17841793, 2009.
[7] Y. H. Zeng and Y.-C. Liang, Spectrum sensing for cognitive
radio based on statistical covariance, IEEE Trans. Veh. Trechnol.,
vol. 58, pp. 18041815, May 2009.
Natonal Conference on Signal & Image Processing -2014 (SIP) 18
Abstract An RTD-based microwave on-off mode oscillator for OOK (On-off keying)
transceiver applications operating at extremely low power consumption is proposed. In
order to achieve the low power operation, the NDC (negative differential conductance)
characteristic at a low voltage arising from quantum effects of the RTD is used for RF
signal generation. The fabricated integrated circuit of the RTD-based oscillator by using an
InP-based RTD/HBT quantum-effect IC technology shows low power consumption of 1.14
mW at an oscillation frequency of 5.725 GHz ISM (Industrial, Scientific, and Medical)
band. The fabricated RTD-based oscillator operates in an on-off mode with a high data
rate of 500 Mb/s. The obtained energy efficiency of 2.28 pJ/bit is found to be the best
reported up to date.
RTD achieved at a low applied voltage from based on-off mode oscillator, which is an
the resonant tunneling nano-quantum well essential component of the wireless OOK
transmitter applications. The proposed on-off
mode oscillator has been implemented by using
an RTD/HBT MMIC (monolithic microwave
integrated circuit) technology and the on-off
mode operation of the implemented IC has been
characterized by on-wafer microwave test
setups.
The detailed layer structure and the fabrication the fully on-mode operation of the oscillator
process have been described elsewhere [4]. biased at a high state of 0.4 V at the Data node
Figure 3 shows the microphotograph of the was first measured on wafer using a N9030A
fabricated oscillator with a chip area of 520480 PXA signal analyzer with a system impedance
m2 excluding the pads. As shown in the of 50 at room temperature.
measured dc characteristics of the fabricated
RTD with an emitter area of 1.0 1.0 m2 As shown in Figure 4, the output spectrum of
(Figure 2), the RTD shows a peak voltage (VP) the oscillator operating in the fully on-mode
of 0.31 V and a peak current (IP) of 850 A with shows an oscillation frequency of 5.725 GHz.
a relatively high PVCR of 9.4 at room The measured phase noise of the oscillator was
temperature. The RTD exhibits the negative obtained to be -104 dBc/Hz at 1 MHz offset.
conductance characteristics in a voltage range The RF power of -13.29 dBm was obtained at
from 0.31 to 0.70 V. The fabricated 1.5 4.0 the output port of the oscillator under the supply
m2 HBT shows a maximum dc current gain of voltage of 0.40 V at the Data node with an RTD
50, and the maximum fT and fMax are 100 GHz bias current of 350
and 100 GHz, respectively. As for the LC power consumption of the oscillator was 0.14
resonator, the inductance of the used spiral mW excluding the output buffer. The power loss
inductor is 2.04 nH. The varactor diode has been in the emitter follower buffer was 1.42 dB. The
fabricated using the base-collector junction of DC-to-RF power efficiency of the oscillator core
the InP-based HBT with a junction size of 29 was found to be 46.4 %. The total power
29 m2. The capacitance of varactor diode is consumption of the RTD oscillator including the
measured to be 280 fF at a reverse bias of 1.6 V. emitter follower output buffer was measured to
be 1.14 mW. The obtained low power
III. MEASUREMENT AND characteristics of the fabricated oscillator are
DISCUSSION mainly originated from the excellent NDC
characteristics of the RTD obtained at the low
Natonal Conference on Signal & Image Processing -2014 (SIP) 22
bias voltage and current condition. In order to oscillator operates at extremely low power
characterize the on-off mode operation, the time- consumption with higher energy efficiency. The
domain measurement for the fabricated RTD extremely low power consumption with the
oscillator was conducted. The NRZ (non return excellent energy efficiency, which is at least 10
to zero)-mode digital input data signal was times better than the other transistor-based
generated by using two pulse pattern generators approaches, indicates that the quantum-effect
(Anritsu MP1763B). The input voltages for the RTD oscillator is very favorable for the low-
on and off states of the input patterns are 0.4 V power short-range wireless transmitters,
and 0 V, respectively. Figure 5 shows the especially such as biomedical applications [15].
measured time-domain waveforms from the The relatively low output power characteristic of
modulated output of the RTD on-off mode the RTD oscillator can be improved by a proper
oscillator at OOK data rates of 250 and 500 design optimization. Because the output power
Mb/s, monitored by an Agilent 83484A digital of RTD oscillator is proportional to the RTD
sampling oscilloscope. The obtained rising time Area, the higher output power can be achieved
(on time) and falling time (off time) of the on- by increasing the RTD area. In this case, since
off mode oscillator are both very fast less than 1 the power consumption of RTD oscillator also
ns, demonstrating the high data rate capabilities increases, the RTD area needs to be carefully
of the RTD oscillator. The obtained high data optimized.
rate performance is due to the inherent high
speed switching capability as well as the high IV. CONCLUSION
NDC arising from the strong quantum-effect- A low power RTD on-off mode oscillator
related I-V nonlinearity of the RTD. The RF operating at the ISM band for OOK transmitter
spectrum efficiency needs to be further applications has been proposed and
characterized and optimized to satisfy the demonstrated based on an InP RTD/HBT MMIC
regulation on RF emission for practical use of technology. The fabricated oscillator showed
the RTD on-off mode oscillator [16]. To extremely low power consumption of 1.14 mW
evaluate the performances of the on-off mode at an oscillation frequency of 5.725 GHz. In
oscillator for OOK transmitter applications, the addition, the fast on-off modulation of the
widely used FOM (figure of merit) for the oscillator has been achieved due to the inherent
energy efficiency is used, which is given by high speed capability and strong NDC
FOM (energy efficiency) = dc power characteristic of the RTD. As a result, the high
consumption / data rate. The energy efficiency energy efficiency of 2.28 pJ/bit at a 500 Mb/s
in a unit of Joule/bit represents the average data rate was achieved. According to the results
amount of energy required to transmit a single obtained in this work, the quantum-effect RTD-
bit of data. The obtained FOM from the based oscillator is very promising for the low
fabricated on-off mode oscillator in this work is power/low energy wireless transmitter
2.28 pJ/bit including the power consumption at applications.
the buffer stage. Table I shows the performance
comparison of the fabricated on-off mode RTD
oscillator with other low-power OOK transmitter
ICs reported in the ISM band frequency range. REFERENCES
The results show that, compared to the
[1] J. Ryu, M. Kim, J. Lee, B. Kim, M. Lee, and
conventional transistor-based OOK transmitters,
S. Nam, Low power OOK transmitter for
the fabricated proposed RTD on-off mode
Natonal Conference on Signal & Image Processing -2014 (SIP) 23
[7] Y. Jeong, S. Choi, and K. Yang, A Sub-100 [15] E. Y. Chow, C.-L. Yang, A. Chlebowski,
W Ku-band RTD VCO for extremely low S. Moon, W. J. Chappell, and P. P. Irazoqui,
power applications, IEEE Microw. Wireless Implantable wireless telemetry boards for In
Compon. Lett., vol. 19, no. 9, pp. 569571, Sep. Vivo transocular transmission, IEEE Trans.
2009. Microw. Theory Tech., vol. 56, no. 12, pp.
32003208, Dec. 2008.
[8] Y. Jeong, S. Choi, and K. Yang, Novel
antiphase-coupled RTD microwave oscillator
Natonal Conference on Signal & Image Processing -2014 (SIP) 24
Kerala, India
e-mail: jayeshg1988@gmail.com
Abstract- The Digital Video Broadcasting second increased with the recent changes in television formats.
generation (DVB T2) system transmitter and These services are at present already used for delivery
receiver are to be implemented and simulated over Digital Video Broadcasting-Satellite (DVBS) and
over cable. But to achieve it overDVB-T is difficult. The
using MATLAB. The existing system uses the
governments and the industries can get attractive
Multiple Input Single Output processing unit
possibilities with the usage of new HDTV services and
(MISO) technology as a means of transmission. migration from analog to digital. In order to increase the
The proposing system is the implementation of performance of DVB-T services, new opportunities have
the DVB T2 system with the support of been provided by the DVB-T2 along with changes in
multiple antenna transmission and multiple channel coding and encoding compression. Hence
antenna reception. Multiple Input Multiple without increasing the radio bandwidth HDTV services
OutputOrthogonal Frequency Division can be provided. There has been a wide interest all
Multiplexing (MIMO-OFDM) is implemented in around the world for HDTV services and before now the
services have been planned and implemented. The
order to support the multiple antenna
migration of encoding of video streams, thechange of
transmission and reception. Here MISO
multiplexing techniques and the implementation of the
processing unit has been replaced by the MIMO new mechanisms in the radio layer is essential in order to
processing unit. In DVB-T2 system transmitter make the services to succeed and also to deliver more
section, reduce the peak to average power ratio digital bandwidth and better forward error recovery. The
to make transmission cheaper. DVBT2 also Digital Video Broadcasting Terrestrial second
supports service specific robustness levels so generation (DVB-T2) uses the Orthogonal Frequency
that both fixed and portable devices will be Division Multiplexing(OFDM) modulation which is
used. MIMO systems can deliver both higher similar to the DVB-T system. A toolkit is provided with
different numbers of carrier such as 1k, 2k, 4k, 8k, 16k,
data throughput and greater robustness by
32k, 16k and 32k extended modulation constellations
taking advantage of the additional signal paths
such as Quadrature Phase Shift Keying (QPSK), 16-
between a transmitter and a receiver. QAM (Quadrature Amplitude Modulation), 64-QAM,
and 256-QAM. Bose-Chaudhuri-Hocquengham (BCH)
Keywords: DVB, Constellation, Alamouti and Low density parity check (LDPC) coding are used by
scheme, MIMO, OFDM. DVB-T2 for the case of error protection. Under certain
conditions in order to provide additional robustness a
new technique has been introduces called as the Rotated
Constellations. DVB-T2 system has proposed with
I.INTRODUCTION
Multiple Input Single Output (MISO) transmission
technology which has multiple transmitting antenna and
Digital Video Broadcasting Terrestrial (DVB T) is the
single receiving antenna.Increasing number of users uses
most popular and successful standard for Digital
the High Definition (HD) television its need better
Terrestrial Television (DTT). For TV broadcasting most
transmission technology that can satisfied the user
of the governments at present are moving to analog
demands. MIMO (Multiple Input Multiple Output) is one
switch off. Even though the digital broadcast standards
of the technologies which can satisfy. In this MIMO
have been available for many years for both terrestrial as
transmission multiple antennas can be used for both
well as over satellite and cable, there has not been much
transmitting and receiving.
benefits in migration for the users. On the other hand the
demand for High Definition TV (HDTV) has been
Natonal Conference on Signal & Image Processing -2014 (SIP) 26
The aim of this project are implementing and out s0 and transmitter 1 (TX1) sends out s1. In the
simulating the transmitter of DVB T2 system with second time slot, TX0 sends out -s1* and TX1 s0*. Here,
integration of MIMO OFDM and also to reduce the * denotes the complex conjugate. The Alamouti scheme
Peak to Average Power Ratio. The MATLAB simulator is is a full rate transmission scheme as one unique symbol
going to be used for implementation and simulation of is transmitted in each time slot.
DVB T2 system with MIMO OFDM.
Tx1
processor is not a part of the T2 system but it contains the B. PEAK TO AVERAGE POWER RATIO
Service splitter or the demultiplexer for separating the (PAPR) REDUCTION
services of the transport streams into system inputs for
the T3. These are then passed on to the individual PLPs In order to decrease the PAPR, two
(Physical Layer Pipes).The total input data capacity of modifications are allowed for the transmitted OFDM
one T2 frame over its duration should not exceed the symbol. The two techniques which are used are the
total available T2 data capacity. Active Constellation Extension Technique and the Tone
Reservation Technique. Both can be used simultaneously.
These techniques are used on the active portion of the
OFDM symbol excluding P1, and after this the guard
A.MIMO PROCESSING intervals are inserted. When both the techniques are
being used simultaneously, the Active Constellation
Extension Technique will be used in the first place to the
signal. But this cannot be used on the Pilot carriers or the
reserved tones or even when the rotated constellations are
being used.
3. DVB-T2 RECEIVER
C.V.Blast Receiver
V-BLAST is abbreviated as the Vertical Bell Labs Space
Time Architecture. It works by increasing the
computational capability of the receiver comparatively to
the minimum mean squarereceiver andzeroforcing
receivers. But at the same time it provides a thorough
signal separation and the capability of tolerance towards
noise. It operates by separating the signals iteratively by
order of strength and finally when all the signals have
been detected and separated it reconstructs the signals[2].
D.Maximum LikelihoodReceiver
This receiver has a better error rate performance than the
remaining receivers, but it is as equally complex in its
architecture.
Natonal Conference on Signal & Image Processing -2014 (SIP) 28
supports different modulation techniques. During this Pilots Arrangement for OFDM Systemover Time Varying
implementation the transmission of random binary ChannelJournal of Networks, Vol. 5, No. 7
stream using the 16-QAM the constellation received [3] Jokela. T,Tupala. M, Paavola. J (2010), Analysis of
from the transmitter is accurate which proves that the Physical Layer Signaling in DVB T2 Systems IEEE
transmission method is going on the right way.It is Transactions on Broadcasting,Vol .56, Issue.3
concluded that the MIMO technology will provides [4] Lukasz Kondrad, Vinod Kumar, Imed Bouazizi,
better solution for large number of users using HDTV. MiikaTupala,And MoncefGabbouj (2010) Cross-Layer
The obtained BER vs. SNR graph shows the accuracy of Optimization of DVBT2 System forMobile
the system after the execution. ServicesInternational Journal of Digital Multimedia
Broadcasting, Volume 2010, Article ID 435405
[5] Robert J. Barsanti, James Larue, Ph.D.(2011), Peak
To Average Power Ratio Reduction For Digital Video
Broadcast T2, IEEE Conference paper.
REFERENCES [6] A. J. Paulraj and H. Bolcskei, Multiple-Input
[1] Jong-SooSeo(2011), Improved CIRBased Receiver Multiple-Output (MIMO) Wireless Systems,The
Design for DVB-T2 System in Large Delay Spread CommunicationsHandbook, CRC Press, second edition,
Channels:Synchronization and EqualizationVol. 57, No. 2002.
1
[2] Hala M. Mahmoud, AllamS. Mousa, Rashid
Saleem(2010), ChannelEstimation Based in Comb-Type
Natonal Conference on Signal & Image Processing -2014 (SIP) 30
The MFCC process consists in taking an entire speech Mel-filter bank is designed to band pass filtering occurring in
utterance as an input, producing a set of acoustic vectors; each auditory system such that it is linear up to 1 kHz and it is
of them having a dimensionality fixed by a number of cepstral logarithmic at higher frequencies. Such a model allows a
th
coefficients (usually 13). The 0 cepstral coefficient is not constant bandwidth and constant spacing on the Mel-frequency
generally used in the cepstral feature vector because it scale and exploits the fact that the speech signal is stationary
represents the energy of the frame[5]. Block diagram of MFCC for short periods of time. It is modelled by constructing the
extraction technique is shown in fig. 1. required number of triangular band-pass filters with 50%
overlap.
The frame-blocking process consists of separating the
speech signal into frames of N samples. The second frame starts B. Cepstral Coefficient Extraction
M samples after the first, with M N. Consequently, it overlaps The spectral properties can be observed efficiently when using
the first with a range of N M samples. The third frame the speech spectrum. First, it provides the Mel-spectrum
overlaps the second with the same number of samples, and so coefficients. Taking the logarithm of the Mel-spectrum
on until the process reaches the end of the speech signal. coefficients, we obtain real numbers, which can therefore be
Following the frame blocking process, the speech signal converted into the cepstral(quefrency) domain. The last step of
encounters discontinuities at the edges of each frame. The MFCC process is converting the mel spectrum to the cepstral
windowing process consists in reducing the signal domain. In order to perform this, we use the Inverse Discrete
discontinuities by lowering down its value to zero at the edges Cosine Transform (IDCT), which will output the Mel
of each frame. For this the hamming window of 20ms width frequency cepstrum coefficients (MFCC). The basic MFCCs
and 10ms shift is used. Before performing the FFT, it is can be upgraded by adding the first and second derivative
necessary to remove frames corresponding to silence, since coefficients to the MFCCs set. Delta- cepstral features are used
these would affect the training and testing steps. The frames
to add dynamic information to the static cepstral features. This
that have energy less than some percentage of maximum of
operation increases the dimension of the feature vectors
frame energy will be eliminated. Upto two consecutive silent
resulting in a higher demand of training data. The first
frames will be retained. If the number of frames as per the
derivative is called delta coefficients and second derivative is
above condition is more than two, then those frames will be
rejected. The following step is used to converting each frame called double delta coefficients[6].
into the frequency domain. The corresponding operation is III. SPEAKER MODELLING USING GMM-UBM
called Discrete Fourier Transform (DFT), which uses the
FFT(Fast Fourier Transform) algorithm. GMM is a probabilistic model that assumes all the data points
are generated from a mixture of a finite number of Gaussian
A. Mel-Frequency Warping (Mel Filter Bank) distributions with unknown parameters. The GMM algorithm
takes a sequence of vectors provided by the MFCC as input and
The spectrum obtained after FFT will have a lot of
uses it to create one model per speaker, which is called the
fluctuations, and we are usually not interested in all the details
Gaussian mixture model. The D-dimensional(here, 39D)
in them. Only the envelope of the spectrum is of interest. A
MFCC is used to obtain the GMM. The idea of using UBM is
filter bank is a series of band pass frequency filters which are
to capture the general characteristics of a population and it is
multiplied one by one with the spectrum in order to get the
output. The filter bank is defined by the shape of the filters and speaker independent. Individual speakers model is obtained by
by their frequency localization (left frequency, central adapting this UBM. For obtaining the UBM, first step is to pool
frequency and right frequency). Filters can be triangular or have the feature vectors of the speakers into one feature matrix.
other shapes and they can be differently located on the Then the next step is to construct the UBM and training it. The
frequency scale[6]. This scale is an auditory scale which is Gaussian mixture density is the sum of M (number of
similar to the frequency scale of the human ear. The mixtures)weighted component densities, given by the following
localization of the central frequencies of the filters is given by equation:
M
f mel
f
2595 * log(1 lin ) (1) p( x) pi bi ( x) (2)
700 i 1
1 1
bi ( x) exp ( x i ) T 1 ( x i )
2 i 2
D/2 1/ 2
(3)
p(i x
B. Testing
t , ) xt
i t 1
T
(4)
In testing the adapted UBM model is used as hypothesized
speaker model. When a speaker claims an identity, it is
p(i x , )
t 1
t necessary to check whether his/her claim is correct or not. For
the verification, the feature vector of the test speaker is
obtained. Then, the log likelihood ratio is obtained. This
Variances: value is compared with a threshold value and make the
T decision, whether the speaker is true or not.
p(i x , ) x t
2
t
i2 t 1
T
i2 (5)
p(i x , )
t 1
t
Mixture weights:
1 T
pi p(i xt , )
T t 1
(6)
Fig. 3. Block Diagram Showing Testing Procedure
1) UBM Adaptation: The UBM helps to improve the The decision making process is pictorically represented in
computational speed of the system. In this paper, only mean of fig. 3. The log likelihood ratio can be computed as:
the mixture densities are used in the adaptation process. The
process is such that, when a speaker enrolls into the system, the ( X ) log p( X hyp ) log p( X ubm ) (12)
UBM is adapted with the speaker dependent features from the
The decision is taken on the basis of the threshold set to the
new speaker. Then the adapted UBM is used as the target
speaker. Assume that the threshold value is . Then the decision
speaker model. The UBM adaptation procedure is explained
is as follows:
below:
If then the speaker will be accepted.
Consider the UBM and training vectors from hypothesized
speaker X {x1 , x2 , x3, ..., xT } . First step is to determine If < then the speaker will be rejected.
the probabilistic alignment of training vectors to the UBM
mixture components. For mixture i in the UBM, we compute IV.EXPERIMENTS
A. Database
pi bi ( xt )
p(i xt ) M
(7) In this work, database composed of 6 speakers uttering
literature in Malayalam language. The database is divided into
p b (x )
k 1
k k t two: 10 minutes speech utterance from each speaker is used for
training and 5 minutes speech utterance from each speaker is
used for testing.
Now the above result and x t is used to compute the new
mean. The new mean can be obtained using the following
equation:
Natonal Conference on Signal & Image Processing -2014 (SIP) 33
% correctness =85.5
Segment 2 V.CONCLUSION
The Gaussian mixture model is fairly effective in the field
of text-independent speaker recognition. This verification
system used MFCC as feature, which uses filter bank with 32
filters. The GMM-UBM is designed for modelling the features.
The percentage is calculated as follows: The UBM adaptation is performed to improve the
computational speed. Then the accuracy of the system is
measured. From the results above shown the system provides
C an accuracy of 85.4 percent. Future work will focus on
Percentage Correctness = 100 (13) applying channel compensation techniques.
N
REFERENCES
where N is the total number of sets of segments, C is the
[1] Douglas A. Reynolds, Thomas F. Quatieri, and Robert B. Dunn,
number of correctly verified segments. Speaker Verification Using Adapted Gaussian Mixture Models,
IDEAL Publications,2000 pp. 19-41, vol. 10
C. Results [2] D. C. Popescu, Kalman Filtering of Colored Noise for Speech
Enhancement, ICASSP98 pp. 997-1000
The number of experiments that are correctly verified when
the test and hypothesized speaker are same is shown in table 1, [3] M. J. F. Gales, speech recognition using parallel model
then table 2 shows the number of experiments that produce the combination, IEEE Trans. Speech Audio Processing, vol. 02,
result that the speaker is an imposter when the test and 1996, pp. 997-1000
hypothesized speaker are different. Then the overall efficiency [4] Dunn R.B, Reynolds D A, and Quatieri T.F, Approaches to
of the system by using the tabulated data is 85.4 %. Speaker Detection and Tracking in Conversational Speech,
Digital Signal Process, vol. 10, 2000, pp. 93-112.
[5] A. R. Douglas, Speaker Identification and verification using
TABLE I.VERIFICATION RESULTS FOR SAME HYPOTHESIZED AND Gaussian mixture speaker models, Speech Communication, vol.
TEST SPEAKER
17, no. 1, 1995, pp. 91-108.
Test Number of correctly verified
segments
[6] Ravi Kumar K M, and Ganesan S, Comparison of
S1 3402 Multidimensional MFCC Feature Vectors for Objective
S2 3410 Assessment of Stuttered Disfluencies , Int. J. Advanced
S3 3425 Networking and Applications , vol. 02, 2011, pp. 854-860.
S4 3422 [7] Tapas Kanungo, Nathan S. Netanyahu, Christine D. Piatko, Ruth
S5 3408
Silverman, and Angela Y. Wu, An Efficient K-Means Clustering
S6 3413
Algorithm: Analysis and Implementation, IEEE Transactions On
% correctness =85.3 Pattern Analysis AND Machine Intelligence, vol. 24, July 2002,
pp. 881-891.
[8] Todd K. Moon, The Expectation-Maximization Algorithm,IEEE
Signal Processing Magazine, Nov 1996, pp. 47-60.
Natonal Conference on Signal & Image Processing -2014 (SIP) 34
AbstractIn this paper, an undecimated double density transactions per pixel. A new method for resolution enhance-
wavelet based image resolution enhancement technique is pro- ment, which preserves high frequency contents of the image
posed. The critically sampled discrete wavelet transform (DWT) is suggested in the paper. Spatial domain techniques lag in
suffers from the drawbacks of being shift-variant and lacking the extraction and preservation of high frequency components
the capacity to process directional information in images. The of an image. This suggests that some other technique not
double density wavelet transform (DDWT) is an approximately
shift-invariant transform capturing directional information. The
involving spatial domain is to be used. So the image needs
undecimated double density wavelet transform (UDWT) is an to be transformed to some other domain, processed and then
improvement of the DD-WT, making it exactly shift-invariant. converted back to the spatial domain. The domain can be
The method uses a forward and inverse (UDWT) to construct a Fourier domain, wavelet domain or any other. Fourier domain
high-resolution (HR) image from the given low-resolution (LR) is more suitable for spectral filtering. The spectral filtering re-
image. The HR image is reconstructed from the LR image using moves particular frequencies from the image. Wavelet domain
the inverse UDDWT. Results are presented and discussed on very separates components of an image into individual matrices.
HR Quick Bird data, through comparisons between state-of-the- These matrices, then can be processed separately and combined
art resolution enhancement method. together to get the desired result.
KeywordsUndecimated double density wavelet transform; Fast algorithms for implementation of discrete wavelet
image resolution; stationary wavelet; resolution enhancement
transform have enhanced the use of the wavelet domain for
image resolution improvement. Different image processing
I. I NTRODUCTION algorithms can be implemented with discrete wavelet transform
Image resolution enhancement is a usable pre-process for (DWT)[3]. DDWT[8] decomposes an image into nine sub
many satellite image processing applications[1], such as bridge bands. These sub bands are of half the dimensions of that
recognition, building recognition and vehicle recognition. Im- of image under consideration. UDDWT[2] is also being used
age resolution enhancement techniques can be categorized into for the image resolution enhancement. UDDWT also has nine
two major types according to the domain that they are applied sub bands similar to DDWT but sub bands in UDDWT are
in: 1) image domain and 2) transform domain. The techniques of same size of that of the image. This paper proposes a new
in the image domain use the statistical and geometric data method for image resolution enhancement, which is based on
directly extracted from the input image itself, while transform- UDDWT.
domain techniques use transformations such as decimated dis-
crete wavelet transform (DWT) to achieve the image resolution II. D EVELOPMENT OF UNDECIMATED DOUBLE DENSITY
enhancement[4]. The decimated DWT has been widely used WAVELET TRANSFORM
for performing image resolution enhancement [6]. A common
Although the Discrete Wavelet Transform (DWT) [7, 8,
assumption of DWT-based image resolution enhancement is
9] is a powerful signal processing tool, it has two severe
that the low-resolution (LR) image is the low-pass-filtered
disadvantages:
subband of the wavelet-transformed high-resolution (HR) im-
age[5]. 1) Lack of shift-invariance, which means that minor
The image resolution is always a key feature for all kinds shifts in the input signal, can cause major variations
of images. With ever increasing size of the displays need in the distribution of energy between wavelet coeffi-
for super resolution images has also been increased. This is cients at different scales.
also impacted by the limited size of the digital image sensor. 2) Since the wavelet filters are separable and real, it
Though widespread commercial cameras provide very high causes poor directional selectivity for diagonal fea-
resolution images, generally the scientific cameras still have tures.
the resolution of only 512 512. Resolution enhancement
has always been associated with the interpolation techniques. The DWT is shift-variant because, the transform coeffi-
Research suggests that interpolation methods increase the cients behave unpredictably under shifts of input signal, a
intensity of low frequency components[9]. This means the problem that has been treated by introducing large amounts
interpolated image will have less number of sharp intensity of redundancy into the transform to make it shift-invariant.
Natonal Conference on Signal & Image Processing -2014 (SIP) 35
R EFERENCES
[1] C. G Hasan Demirel and Gholamreza Anbarjafari, Discrete
Wavelet Transform-Based Satellite Image Resolution Enhance-
ment , IEEE Transactions on Geoscience and Remote sensing,
June 2011, vol. 49
[2] J. E. Fowler, The redundant discrete wavelet transform and
Fig. 5. PSNR values for different resolution enhancement techniques from
256x256 to 512x512
additive noise , IEEE Signal Processing Letters, 2005, vol.
629-632
Natonal Conference on Signal & Image Processing -2014 (SIP) 37
Abstract A category of new microstrip patch antennas with system and they are the connecting links between the
very low profile, wide bandwidth and, most importantly, vertical transmitter and free space or free space and the receiver. Thus
polarization are analysed. A monopolar microstrip circular patch antennas play very important role in finding the characteristics
antenna that has a monopole like radiation pattern is studied. of the system in which antennas are employed. Antennas are
The monopolar patch antenna can produce a conical radiation employed in different systems in different forms. That is, in
pattern in the elevation pattern and an omnidirectional pattern
some systems the high gain of the antenna is considered,or the
in the horizontal plane, like a monopole antenna. The monopolar
circular patch antenna has a potential to replace the conventional directional properties,or the omnidirectional radiation pattern
monopole antenna for the circumstances that require an antenna ,or the high beam width is considered.That is the selection of
with very low profile and vertical polarization. A microstrip the antenna depends on its application.
rectangular patch antenna with high gain and vertically
polarized radiation at endfires is also considered. The II. POLARIZATION IN ANTENNAS
rectangular patch antenna produces a monopole like radiation
pattern in the main elevation plane and an 8-shaped radiation A.HORIZONTAL POLARIZATION
pattern in the horizontal plane. The Yagi antenna produces a
main beam radiating very close to forward endfire and a vertical
polarization at the endfire. Horizontally polarized antennas have their electric
field parallel to the Earths surface. Horizontally polarized
Index Termsvertical polarization and horizontal polarization, signals oscillate from left to right. Horizontal polarization
microstrip patch, shorting bias,modes of antenna,endfire frequencies are parallel to and touch the earth. Since the earth
radiation acts as a good conductor at low frequencies, it shorts some of
I.INTRODUCTION the frequencies and prevents the signals from traveling very
far.
As the communication devices are becoming smaller
due to greater integration of electronics, the antenna becomes
a significantly larger part of the overall package volume. This
results in a demand for similar reductions in antenna size. B.VERTICAL POLARIZATION
When designed at lower microwave frequency spectrum,the
size of a conventional microstrip antenna is somewhat large. An antenna is vertically polarized when its electric
Sometimes the size of the antenna even exceeds the dimension field is perpendicular to the Earths surface. Vertically
of the repeater or receiver system and thus is unsuitable for polarized signals oscillate from top to bottom. Signals are
mounting conformably on the existing repeater/receiver transmitted in all directions. Therefore vertical polarization is
system. For many antenna applications, such as handheld used for ground-wave transmission, allowing the radio wave
transceivers and handheld computers, small size is extremely to travel a considerable distance along the ground surface with
important. For fixed wireless applications also, the small sized minimum attenuation. For the electromagnetic wave the
antennas plays an important role. The new trends in antenna polarization is effectively the plane in which the electric wave
design mainly focuses on the compactness of antenna,its vibrates. This is important when looking at antennas because
robustness and integration with the existing RF circuit they are sensitive to polarisation, and generally only receive or
components. Antennas are basic components of any electric transmit a signal with a particular polarization.
Natonal Conference on Signal & Image Processing -2014 (SIP) 41
transmission line in Fig. 2(b) represents the long microstrip Most RF antenna designs are operated around the
line, and the shorting elements at both ends represent the resonant point. This means that there is only a limited
closely-spaced shorting-vias at both ends of the microstrip bandwidth over which an RF antenna design can operate
line. Suppose the antenna works in the mode with respect to efficiently. Outside this the levels of reactance rise to levels
the direction. In other words, the length of the transmission that may be too high for satisfactory operation. Other
line is a half wavelength of the waves in the direction. characteristics of the antenna may also be impaired away from
the centre operating frequency.
The antenna bandwidth is particularly important
where radio transmitters are concerned as damage may ccur to
the transmitter if the antenna is operated outside its operating
range and the radio transmitter is not adequately protected. In
addition to this the signal radiated by the RF antenna may be
less for a number of reasons.
For receiving purposes the performance of the
antenna is less critical in some respects. It can be operated
outside its normal bandwidth without any fear of damage to
the set. Even a random length of wire will pick up signals, and
it may be possible to receive several distant stations. However
Fig.2:Shorting bias at the two sides of rectangular antenna for the best reception it is necessary to ensure that the
performance of the RF antenna design is optimum.
V.RECTANGULAR PATCH ANTENNA WITH
Another feature of an antenna that changes with
SHORTING-VIAS ALONG THE LONGITUDINAL
frequency is its radiation pattern. In the case of a beam it is
DIRECTION
particularly noticeable. In particular the front to back ratio will
fall off rapidly outside a given bandwidth, and so will the
In order to have a wider bandwidth for the long
gain. In an antenna such as a Yagi this is caused by a
rectangular patch, another group of shorting-vias are
reduction in the currents in the parasitic elements as the
introduced alining along the longitudinal direction, as shown
frequency of operation is moved away from resonance. For
in Fig. 3. The shorting-vias placing at in the longitudinal
beam antennas such as the Yagi the radiation pattern
direction have a period of and a diameter of .When these
bandwidth is defined as the frequency range over which the
shorting-vias are employed and , another resonant mode called
gain of the main lobe is within 1 dB of its maximum.
TM12 is generated in addition to its fundamental mode. By
adjusting the position and the period of the shorting-vias, the
resonant frequency of the mode can be tuned to be close to
that of the TM10 mode, and therefore a wide bandwidth is
obtained due to the coupling of the two modes.
VI.RADIATION PATTERNS
Fig.4:Radiation pattern along the elevation plane
Since the thickness of the substrate of the antenna is
very small compared to the wavelength in free space, the One major feature of an RF antenna that does change
magnetic currents at the apertures of the antenna can be with frequency is its impedance. This in turn can cause the
assumed for both the TM10 and TM12 modes. amount of reflected power to increase. If the antenna is used
for transmitting it may be that beyond a given level of
reflected power damage may be caused to either the
Natonal Conference on Signal & Image Processing -2014 (SIP) 43
transmitter or the feeder, and this is quite likely to be a factor 100 times stronger (compared to an isotropic radiator), will
which limits the operating bandwidth of an antenna. Today also capture 100 times as much energy as the isotropic antenna
most transmitters have some form of SWR protection circuit when used as a receiving antenna. As a consequence of their
that prevents damage by reducing the output power to an directivity, directional antennas also send less (and receive
acceptable level as the levels of reflected power increase. This less) signal from directions other than the main beam. This
in turn means that the efficiency of the station is reduced property may be used to reduce interference.
outside a given bandwidth. As far as receiving is concerned
the impedance changes of the antenna are not as critical as
they will mean that the signal transfer from the antenna itself
to the feeder is reduced and in turn the efficiency will fall. For
amateur operation the frequencies below which a maximum
SWR figure of 1.5:1 is produced is often taken as the
acceptable bandwidth.
VII.RETURN LOSS
VIII.CONCLUSION
When transmitting, a high gain antenna allows more
of the transmitted power to be sent in the direction of the
receiver, increasing the received signal strength. When When a rectangular patch antenna has only two
receiving, a high gain antenna captures more of the signal, columns of shorting-vias at both ends of the patch, it has a
again increasing signal strength. Due to reciprocity, these two bandwidth of 3.8% for a profile of 0.03 wavelengths in free
effects are equal - an antenna that makes a transmitted signal
Natonal Conference on Signal & Image Processing -2014 (SIP) 44
space. The antenna works in its TM12 mode. When another [9] Vivek Kumar Agarwala, Anand Kumar Shawa ,A Novel
group of shorting-vias are introduced along the longitudinal Compact Dual Frequency Microstrip Antenna, Procedia
direction, another resonant mode (TM12 mode) is generated in Technology, vol. 4, pp.427 430, 2012
addition to the TM01 mode, and the bandwidth is improved
greatly. A monopole like radiation pattern with vertical
polarization at endfires is produced in the main elevation
plane. When the length of the rectangular microstrip increases,
its maximum directivity is increased due to its larger radiation
apertures, butits bandwidth is narrowed on the other hand.
REFERENCES
[5]T.Shanmuganantham,S.Raghavan,Designofacompactbroa
dbandmicrostrippatchantennawithprobefeeding for wireless
applications,Int. J. Electron. Commun. (AE) , vol. 63
,pp.653659 , 2009
Abstract Robustness of speaker recognition systems is crucial These effects in turn mask higher frequencies in the speech
for real-world applications, which typically contain both spectrum and blur spectral details, both of which are
additive noise and room reverberation. However, the useful acoustic cues for speaker i d e nntt i f i c a t i on [ 1] [ 2] .
combined effects of additive noise and reverberation have
been rarely studied in speaker identification (SID). Room A blind spectral weighting (BSW) technique for
reverberation poses various deleterious effects on performance suppressing the reverberation overlap-masking
overlap effect on
of automatic speech systems. Speaker identifi fication (SID) SID systems is used to avoid these effects
effects.
performance, in particular, degrades rapidly as
reverberation time increases. Reverberation causes two forms
of spectro-temporal distortions on speech signals: i) self-
II. REVERBERATION
masking which is due to early reflections and ii) overlap-
masking which is due to late reverberation. Overlap-masking
Overlap In a reverberant enclosure, sound waves arrive at the
effect of reverberation has been shown to have a greater receiver (e.g., ears or microphone) via a direct path and
adverse impact on performance of speech systems. a blind via multiple paths and directions after reflecting off
spectral weighting (BSW) technique for suppressing the walls and objects defining the acoustic enclosure. This
reverberation overlap-masking effect on SID systems is used to
avoid this effects. The technique is blind in the sense that prior
is illustrated in Fig. 1 where the sound source is
knowledge of neither the anechoic signal nor the room impulse captured at the microphone as a delayed sum of the
response is required. direct sound and its reflections from the wall. The
reflections arriving within 50-80 80 ms after the direct
Index TermsMismatch conditions, overlap- overlap masking
sound are called early reflections, which tend to build
effect, reverberation, speaker verification. up to a level louder than the direct sound and cause an
internal smearing effect known as the self-masking
effect. Echoes reaching the receiver after early
I. INTRODUCTION reflections are called late reflections, which tend to
Automatic speech systems integrated in to a variety of smear the direct sound over time and mask succeeding
electronic/mobile components of an individuals daily life. sounds. This phenomenon is commonly referred to as the
Nevertheless, providing robustness to these systems still overlap-masking effect, and has been shown to be the
remains a challenge because of the variety of acoustic primary cause of degraded speech identification
mismatch scenarios that may occur between training and performance for both human and machine listeners. The
test conditions due to back- ground noise, room overlap-masking effect can also mask/obscure spectral
cent, language,
reverberation, communication channel, accent, details and acoustic cues essential for automatic SID,
emotions, vocal effort, etc. resulting in a major drop in performance
performance.
Robustness of automatic speaker recognition is critical for
real-world applications. In daily acoustic environments,
additive noise, room reverberation and channel/handset
variations conspire to pose considerable
c h a l l e n g e s t o s u c h systems.
Reverberation poses various detrimental effects on
spectro-temporal characteristics of speech signals, most
notably including temporal smearing, filling dips and gaps
in the temporal envelope, increasing the prominence of
Fig. 1. Direct and reflected sound signal components.
low-frequency energy, and flattening formant transitions. denotes t h e source-to-microphone distance.
Natonal Conference on Signal & Image Processing -2014 (SIP) 47
Fig. 2. Sample spectrograms for the sentence she had your dark suit in greasy wash water all year, in anechoic quiet condition (left), and
under reverberant condition.
(9)
VI. CONCLUSION
AbstractDigital watermarking is an important tool for copy- but they are generally fragile to image processing operations
right protection. Discrete wavelet transform (DWT), and singular or other attacks. On the other hand, the transform domain
value decomposition (SVD) are the two commonly used trans- techniques embed the watermark by modulating the magnitude
forms among digital watermarking. We propose a 3 level DWT-
SVD domain watermarking scheme in this paper. Performance of coefficients in a transform domain, such as discrete cosine
of watermarking scheme has been improved for color images transform, discrete wavelet transform (DWT), and singular
in this paper. The performance is enhanced by increasing the value decomposition (SVD). Transform domain methods can
level of DWT in the proposed scheme. This is a robust color embed more information and yields more robustness against
image watermarking method for the copyright protection. After many common attacks, but the computational cost is higher
the third level of decomposition of the host image into 10 sub
bands, we apply SVD to a low frequency subband and embed than spatial domain watermarking methods.
singular values of particular sub band of the watermark into The DWT based watermarking techniques considers the
them. Our scheme has advantages of robustness, large capacity human visual properties and the use of human visual model
and performance improvement by using 3 level DWT and SVD. guarantees the imperceptibility of the watermark. The per-
This paper compares the performance of the algorithms for one formance improvements in DWT based digital image water-
level and three level DWT.
Index TermsWatermarking,Encoding, Decoding, Discrete marking algorithms can be obtained by increasing the level of
wavelet transform (DWT), Singular value decomposition (SVD). DWT. Here 3 level DWT is employed. In the DWT domain
an image is decomposed into four subbands LL, LH, HL, and
HH at level 1 , where LH, HL, and HH represent the finest
I. I NTRODUCTION scale wavelet coefficients and LL represents the coarse level
We are in the midst of digital revolution. The popularity of coefficients. The LL band is selected for decomposition as it
the internet has resulted in the rapid growth of multimedia offers greater robustness.
technology, thus the users get more and more chances to SVD is an optimal matrix decomposition technique in a least
use multimedia data. Thus the digital technologies have new square sense. Its advantage is that it packs maximum energy
threats for the intellectual properties and contents such as into as few coefficients as possible. SVD based watermarking
easy access to information, modification/falsification facil- method embeds data by modifying the singular values. The
ities, redistribution facilities and perfect copies. Therefore proposed watermarking scheme is based on DWT and SVD
protection of intellectual property rights of digital media has and it takes the advantage of both the transform, ie robustness
become an urgent issue. That is, ease by which a digital and increased capacity.
information can be duplicated and distributed has lead to the The rest of the paper is organized as follows. In Section
need for effective copyright protection tools. There comes II, we present the current state of the art related to the image
the role of watermarking. Digital watermarking has various watermarking. In Section III, we discuss the transforms used
applications, such as copyright protection, authentication and for watermarking. In Section IV, we propose the watermarking
secret communication. scheme in 3- level DWT-SVD domain and in Section V we
Watermarking is a process of embedding data called a discuss its experimental results. In section VI we discuss
watermark into a multimedia object as reliable and robust, theperformance evaluation and finally Section VII summarizes
such that watermark can be detected or extracted later to make this paper with some concluding remarks.
an asertion about the object . Watermarking can be either
visible or invisible. Depending on the domain in which the II. C URRENT S TATE OF THE A RT
watermark is inserted, these techniques are classified into two Watermarking is a protection tool for data encoding and
types, i.e., spatial domain and transform domain methods. copyright protection [1]. Digital watermarking method are
In spatial domain method the watermark is embedded into mainly of two types based on processing method, spatial
the spatial domain component of the original image. It has domain, like LSB and transform domain, like DCT, DWT
the advantages of low complexity and easy implementation methods. In the spatial domain method the watermark is
Natonal Conference on Signal & Image Processing -2014 (SIP) 51
embedded in the lower order bit planes .It is an easy im- decomposition is shown in figure 1. The performance in DWT
plementation method and has low complexity but is not based watermaking scheme can be enhanced by increasing the
robust against the common atttacks. Transform domain method level of DWT. The high frequency region gives the information
produce high quality watermarked image by first transforming about the edge components, while the low frequency region
the original image into the frequency domain by the use of is decmposed again into high and low frequency parts. When
Fourier Transform, Discrete Cosine Transform or Discrete watermark is inserted by changing the intermediate frequency
Wavelet transforms and modifying the transform coefficients and low frequency components, the robustness can be highly
to embed the watermark data. It is observed that transform improved. While taking into account the anti interference of
based watermarking is much better than the spatial domain the watermark the high frequency components are used for
watermarking. watermarking since the human eye is less sensitive to changes
Xia, Xiang-Gen and Boncelet, proposed a watermarking in edge regions [6]. The extraction is done by applying inverse
scheme based on the Discrete Wavelet Transform (DWT) [2]. procedure ,ie sequentially from a low resolution to the higher
In this literature the watermark was taken as a Gaussian noise, resolution [2].
and was added to the middle and high frequency bands of the
image. In the extraction process the DWT of a an image was
computed. Sections of the watermark were extracted and was
correlated with the sections of the original watermark. When
the cross correlation was found above a threshold, then the
recovered watermark was correct. If cross correlation value is
less than the threshold, the image was further decomposed into
finer and finer sub bands until the entire extracted watermark
was correlated with the entire original watermark. Performance Fig. 1. 3-Level discrete wavelet decompositions
of this scheme was analysed and results shows that DWT
is more robust to attack than DCT. But the problem with B. Single Value decomposition (SVD)
the proposed scheme is that, this technique is more prone
to geometric attacks. SVD is another transform used for SVD is a transform method used for watermarking. It is a
watermarking, which was applied to the host image, and in this linear algebra approach. SVD is used for various applications
bits of watermark were directly added to the singular values such as solving most linear least-squares problems, computing
of host image in [3]. It was one of the watermarking method pseudo-inverse of a matrix and multivariate analysis. It is
applied for copyright protection. an optimal matrix decomposition method, and accumulates
A hybrid DWT SVD domain watermarking scheme consid- maximum signal energy into as few coefficients as possible.
ering the human visual model has the advantages of robustness Its capable of adapting to local variations of a given image.
for its embedding data into all frequencies and large capacity Also singular value has properties of stability, proportion
for using DWT and SVD [4]. A wavelet domain watermarking invariance, and rotation invariance [7]. SVD has a variety of
method, combined with the human visual system, adaptively image processing applications such as noise estimation [8] and
adjusts the watermark embedding strength, improves the ro- digital watermarking. Also SVD has the ability to adapt with
bustness of watermark, but watermarking capacity of the the variations in local statistics of an image, so watermarking
algorithm is smaller [5]. method using SVD are typically of large capacity. The image
is decomposed by SVD on three matrices and the equation for
III. T RANSFORMS USED FOR PROPOSED WATERMARKING SVD is given by
A. Descrete Wavelet Transform (DWT)
DWT is being used in digital image watermarking more I = U SV T (1)
frequently due to its excellent spatial localization and multi Singular value S, specifies the luminance of an image layer
resolution characteristics, which are similar to the theoretical while the singular vectors U and V specifies the geometry of
models of the human visual system. DWT is a mathematical the image layer.
tool for decomposing an image in a pyramid like structure. .
The transform is based on small waves, called wavelets.
Wavelets are created by translations and dilations of a fixed IV. P ROPOSED WATERMARKING SCHEME IN DWT-SVD
function called mother wavelet. DWT gives the multiresolu- DOMAIN
tion description of an image. In the analysis of multi level The watermarking method comprises of two steps , water-
decomposition of images, it was found that the approximation mark embedding and watermark extracting process.
sub images at the lowest level accumulate most energies of the
image. They area vital part of vision, and they have a great A. Watermark Embedding
visual capacity. So it is the most suitable location to embed The watermark embedding process is shown in figure 2. In
the digital watermark information. The DWT decompose the this approach firstly the host image, H is taken and 2D, 3 level
signal into high and low frequency parts. The 3 level DWT DWT is applied to the image which decomposes image into
Natonal Conference on Signal & Image Processing -2014 (SIP) 52
low frequency and high frequency components. In 2D DWT, values of the watermark, the sub-band LL3 is computed back
for each level of decomposition, there are 4 sub bands, LL1, by the equation
LH1, HL1, and HH1. For each level of further decomposition,
the LL sub band of the previous level is used. ie, to perform T
LL3m = ULL3 SLL3 VLL3 (5)
second level decomposition, the DWT is applied to LL1 band
which decomposes the LL1 band into the four sub bands LL2, The left and right singular vectors of LL3 sub-band and the
LH2, HL2, and HH2. In the third level of decomposition, the modifed matrix SLL3 of singular values is used for inverse
DWT is applied to LL2 band which decompose this band SVD inorder to compute LL3. The result is the transformed
into four sub bands LL3, LH3, HL3, HH3. This results in sub-band LL3m which is used with other sub bands to inverse
10 sub bands per component. LH1, HL1, and HH1 forms DWT. 3-level Inverse DWT is applied to the watermarked
the highest frequency bands in the image, while LL3 is the image coefficient to produce the final secure watermarked
lowest frequency band. The low frequency component LL3 image Hw .
is chosen for watermark insertion as it gives more robustness.
B. Watermark Extraction
Each colour layer of the low frequency subband is decomposed
by means of SVD on three matrices. The equation for SVD is The watermark extraction process is shown in figure 3. The
T
LL3 = ULL3 SLL3 VLL3 (2)
where LL3 is the chosen sub band, the diagonal entries of S
are the singular values, ULL3 is the matrix of the left singular
vectors of LL3, and the columns of VLL3 are the right singular
vectors of LL3. All new matrix has the same dimension as that
of the selected matrix LL3. The important feature is that each
singular value S specifies the luminance of an image layers
while the corresponding pair of singular vectors specifes the
geometry of the image layer .
T
Wm = UWm SWm VW (3) SE = SLL3w SLL3 /h (8)
m
The SW matrix is used for watermarking. The main idea of The extracted singular value matrix is used for inverse
the watermark embedding follows the equation singular value decomposition.
The right and left singular value with extracted singular
value matrix are used for it. The inverse singular value
SLL3 = SLL3 + hSWm (4)
decomposition uses the equation
where h is the scaling factor. After the singular values of
T
host are altered by taking the weighted sum of the singular W0 = Uw SE Vw (9)
Natonal Conference on Signal & Image Processing -2014 (SIP) 53
S2
P SN R = 10 log (10)
M SE
Here S=255 for an 8 bit image. PSNR is basically the SNR
when all pixel values are equal to the maximum possible (a) (b)
value. Mean square error (MSE) is computed by averaging
the squared intensity of the original (input) image and the Fig. 5. Comparison of performance in one level and three level DWT- SVD
resultant (output) imaage pixels as in domain watermarking
Natonal Conference on Signal & Image Processing -2014 (SIP) 54
Abstract Sixth Sense Technology integrates digital arena. This technology has emerged, which has its relation to
information into the physical world and its objects, making the the power of these six senses. Our ordinary computers will
entire world your computer. It is a wearable gestural interface soon be able to sense the different feelings accumulated in the
that augments the physical world around us with digital surroundings and it is all a gift of the Sixth Sense
information and lets us use natural hand gestures to interact with
Technology newly introduced. Sixth Sense is a wearable
that information. It can turn any surface into a touch-screen for
computing, controlled by simple hand gestures. gesture based device that augments the physical world with
digital information and lets people use natural hand gestures to
The Sixth Sense prototype comprises a pocket projector, interact with that information.
a mirror and a camera contained in a pendant like, wearable Right now, we use our devices
device. Both the projector and the camera are connected to a (computers, mobile phones, tablets, etc.) to go into the internet
mobile computing device in the users pocket. The projector and get information that we want. With Sixth Sense we will
projects visual information enabling surfaces, walls and physical use a device no bigger than current cell phones and probably
objects around us to be used as interfaces; while the camera
eventually as small as a button on our shirts to bring the
recognizes and tracks user's hand gestures and physical objects
using computer-vision based techniques. The software program internet to us in order to interact with our world!
processes the video stream data captured by the camera and Sixth Sense will allow us to interact with
tracks the locations of the colored markers (visual tracking our world like never before. We can get information on
fiducially) at the tip of the users fingers anything we want from anywhere within a few moments! We
will not only be able to interact with things on a whole new
Index TermsSixth Sense Technology, Camera, Color level but also with people! One great part of the device is its
Markers,Mobile, Pocket Projector. ability to scan objects or even people and project out
information regarding what you are looking at.
I. INTRODUCTION External devices can also be controlled using hand gestures.
Depending on hand gestures signals are sent to microcontroller.
Weve evolved over millions of years to sense the world
Microcontroller processes the data and transmits the signal to
around us. When we encounter something, someone or some
external device. The receiver at the external device receives the
place, we use our five natural senses which include eye, ear,
transmitted signal. By using a relay, power is given to the
nose, tongue mind and body to perceive information about it; devices which is controlled by the microcontroller depending
that information helps us make decisions and chose the right
on the received signal..
actions to take. But arguably the most useful information that
can help us make the right decision is not naturally perceivable
with our five senses, namely the data, information and
WHAT IS SIXTH SENSE?
knowledge that mankind has accumulated about everything and
which is increasingly all available online.
Sixth Sense in scientific (or non-
Although the miniaturization of
scientific) terms is defined as Extra Sensory Perception or in
computing devices allows us to carry computers in our pockets,
short ESP. It involves the reception of information not gained
keeping us continually connected to the digital world, there is
through any of the five senses. Nor is it taken from any
no link between our digital devices and our interactions with
experiences from the past or known. Sixth Sense aims to more
the physical world. Information is confined traditionally on
seamlessly integrate online information and tech into everyday
paper or digitally on a screen. Sixth Sense bridges this gap,
life. By making available information needed for decision-
bringing intangible, digital information out into the tangible
making beyond what we have access to with our five senses, it
world, and allowing us to interact with this information via
effectively gives users a sixth sense
natural hand gestures. Sixth Sense frees information from its
confines by seamlessly integrating it with reality, and thus
making the entire world your computer.
Sixth Sense Technology, it is the
newest jargon that has proclaimed its presence in the technical
Natonal Conference on Signal & Image Processing -2014 (SIP) 56
II. . EARLIER SIXTH SENSE PROTOTYPE a pendant-like mobile wearable device. Both the projector and
the camera are connected to the mobile computing device in
the users pocket.
We can very well consider the Sixth Sense Technology as a
blend of the computer and the cell phone. It works as the
device associated to it is hanged around the neck of a person
and thus the projection starts by means of the micro projector
attached to the device. Therefore, in course, you turn out to be
a moving computer in yourself and the fingers act like a mouse
and a keyboard.
The prototype was built from an ordinary webcam and a
battery-powered 3M projector, with an attached mirror all
Maes MIT group, which includes seven graduate connected to an internet-enabled mobile phone. The setup,
students, were thinking about how a person could be more which costs less than $350, allows the user to project
integrated into the world around them and access information information from the phone onto any surface walls, the body
without having to do something like take out a phone. They of another person or even your hand.
initially produced a wristband that would read a Radio
Frequency Identification tag to know, for example, which book IV. EXTERNAL DEVICE CONTROL
a user is holding in a store. External devices can also be controlled using same
They also had a ring that used infrared to technology. External device control is advancement from
communicate by beacon to supermarket smart shelves to give existing system. Existing system can only able to gather the
you information about products. As we grab a package of information as well as manipulating the information according
macaroni, the ring would glow red or green to tell us if the to our requirement. This modification to the current system will
product was organic or free of peanut traces whatever criteria expand the scope of Sixth Sense Technology. Using this
method we can control devices like fan, TV, etc. This system
we program into the system.
needs transmitter, receiver, microcontroller, etc. to transmit
They wanted to make information more data from the mobile computing device and to receive the
useful to people in real time with minimal effort in a way that signals for controlling different devices.
doesnt require any behavior changes. The wristband was
getting close, but we still had to take out our cell phone to look V. WORKING OF SIXTH SENSE TECHNOLOGY
at the information. Thats when they struck on the idea of
accessing information from the internet and projecting it. So A. COMPONENTS
someone wearing the wristband could pick up a paperback in
the bookstore and immediately call up reviews about the book, The hardware components are coupled in a
projecting them onto a surface in the store or doing a keyword pendant like mobile wearable device.
search through the book by accessing digitized pages on
Camera
Amazon or Google books.
Projector
They started with a larger projector that was Mirror
mounted on a helmet. But that proved cumbersome if someone Mobile Component
was projecting data onto a wall then turned to speak to friend
Color Markers
the data would project on the friends face .
Transmitter
III. RECENT PROTOTYPE Receiver
Microcontroller
Relays
Information is sent to the Smartphone for processing or she is interacting with it. The software searches the
The downward-facing projector projects the output internet for information that is potentially relevant to that
image on to the mirror situation, and then the projector takes over.
Mirror reflects image on to the desired surface
Thus, digital information is freed from its confines and C. EXTERNAL DEVICE CONTROL
placed in the physical world
Data from M.C Transmitter
Sixth sense device analyses what user sees and Computing device
visually augments the surfaces and physical objects user is
interacting with. The key here is that Sixth Sense recognizes
the objects around us, displaying information automatically and Receiver M.C Relay External
letting us access it in any way we want, in the simplest way Device
possible. The technology is mainly based on hand gesture M.C Micro Controller
recognition, image capturing, processing and manipulation, etc.
External devices can also be controlled using hand gestures.
Depending on hand gestures, signals are sent to
microcontroller. Microcontroller processes the data and
transmits the signal to external device. The receiver at the
external device receives the transmitted signal .By using a relay
power is given to the devices which is controlled by the
microcontroller depending on the received signal.
VI. APPLICATIONS
a) Viewing Map:
project them onto any surfaces and then use gestures to sort g) External Device control:
through those photos and organize and resize them.
External devices can also be controlled using same
c) Drawing Application: technology. Using this method we can control devices like fan,
TV, etc. External device control is advancement from existing
system.
h) Feed information on people
i) Other Applications:
Sixth Sense also lets the user draw icons or symbols in the
We can make calls with the help of Sixth Sense device. The air using the movement of the index finger and recognizes
Sixth Sense device is used to protect the keyboard into your those symbols as interaction instructions. For example,
palm and using that virtual keypad we can make calls to drawing a magnifying glass symbol takes the user to the map
anyone. application or drawing a @ symbol lets the user check his
e) . Interacting with Physical Objects: mail.
The Sixth Sense system also helps to interact with VII. ADVANTAGES AND ENHANCEMENTS
physical objects we use in a better way. It augments physical
objects by projecting more information about these objects Sixth Sense is an user friendly interface which
projected on them. For example, a gesture of drawing a circle integrates digital information into the physical
on the users wrist projects a watch on the users hand. world and its objects, making the entire world your
Similarly a newspaper can show live video news or dynamic computer.
information can be provided on a regular piece of paper. Sixth Sense does not change human habits but
causes computer and other machines to adapt to
human needs.
It uses hand gestures to interact with digital
information.
Supports multi-touch and multi-user interaction
Data access directly from machine in real time
It is an open source and cost effective and we can
mind map the idea anywhere
f) Flight Updates: It is gesture-controlled wearable computing device
that feeds our relevant information and turns any
The system will recognize your boarding pass and let you surface into an interactive display.
know whether your flight is on time and if the gate has It is portable and easy to carry as we can wear it in
changed. our neck.
The device could be used by anyone without even a
basic knowledge of a keyboard or mouse.
There is no need to carry a camera anymore. If we
are going for a holiday, then from now on wards it
will be easy to capture photos by using mere
fingers
Natonal Conference on Signal & Image Processing -2014 (SIP) 59
REFERENCES
VIII. FUTURE ENHANCEMENTS
S. Sadhana Rao, Sixth Sense Technology, Proceedings of the
To get rid of color markers International Conference on Communication and Computational
To incorporate camera and projector inside mobile Intelligence 2010, Kongu Engineering College, Perundurai,
Erode, T.N.,India.27 29 December,2010.pp.336-339.
computing device.
P. Mistry, P. Maes. SixthSense A Wearable Gestural
Whenever we place pendant- style wearable device on Interface. In the Proceedings of SIGGRAPH Asia 2009, Sketch.
table, it should allow us to use the table as multi touch Yokohama, Japan. 2009
user interface. P. Mistry, P. Maes, L. Chang. WUW - Wear Ur World - A
Applying this technology in various interest like Wearable Gestural Interface. In the CHI '09 extended abstracts
gaming, education systems etc. on Human factors in computing systems. Boston, USA. 2009
To have 3D gesture tracking. http://www.pranavmistry.com
To make sixth sense work as fifth sense for disabled http://www.pranavmistry.com/projects/sixthsense/
person.
CONCLUSION
AbstractCognitive Radio (CR)is viewed as a promising Robust against noise uncertainity and works well for OFDM
approach to improving the utilization of a precious natural signals with better performance.For a wide frequency band,
resource:the radio electromagnetic spectrum.It is intended to this method can estimate spectrum holes in the signal spectrum
coexist with primary users (PUs) for using the underutilized with simple structure and low computational complexity.
spectrum without any harmful interference. Spectrum Sensing is
the Key function of Cognitive Radio for detecting primary users.
Wavelet Packet Entropy based sensing is a best technique offering II. C OGNITIVE R ADIO
robustness towards noise uncertainty and low computational Cognitive Radio (CR) is an intelligent wireless
complexity compared to conventional schemes. communication system that is cognizant (hence the name) of
In this Paper a Wavelet Packet Entropy based Algorithm, for
its environment, learns from it and adapts its transmission
detection of primary users in Cognitive Radio networks is
extended to OFDM based Cognitive Radio.It is used for sensing features according to statistical variations in the environment
the presence of primary user within OFDM band. Results to maximize utilization of premium resources such as
show that method offer accurate detection and position location spectrum while ensuring good QoS.
information of primary signal in OFDM band. Spectral masking
and reshaping can be carried out accordingly. The idea of CR was first proposed by Joseph Mitola III
KeywordsCognitive Radio (CR); Primary User and Gerald Q. Maguire.
(PU); Wavelet Packet Entropy(WPE); Wavelet Packet
Transform(WPT);Spectrum Sensing Cognitive radio can be defind as: It is an intelligent wireless
communication system that is aware of its surrounding
I. I NTRODUCTION environment (i.e., outside world), and uses the methodology
of understanding-by-building to learn from the environment
In the last decade, as wireless communications have be- and adapt its internal states to statistical variations in the
come widely popular, the seriousness of spectrum scarcity incoming RF stimuli by making corresponding changes in
is being on the rise gradually. On one hand, the use of certain operating parameters (e.g., transmit-power, carrier-
radio spectrum is licensed by governments, resulting in the frequency, and modulation strategy) in real-time, with two
unallocated spectrum segments become extremely rare. On the primary objectives in mind:
other hand, recent studies of the spectrum usage indicate that
most of the spectrum is not used for most of the time[3] . highly reliable communications whenever and wher-
To utilize the radio resources more efficiently, cognitive ever needed
radio (CR) technology has been valued increasingly as a
efficient utilization of the radio spectrum.
potential candidate. CR is defined as having the capability of
sensing underutilized spectrum holes, and communication by A CR network typically involves two types of users: primary
using these unused spectrum resources without causing harm- users (PUs), who are incumbent licensed users of the spectrum,
ful interference to the primary users (PU). Spectrum utilization and CR users (also known as secondary users),who try to
can be improved significantly by making it possible for a opportunistically access the unused licensed spectrum as long
secondary user (SU) to access a spectrum hole unoccupied. as harmful interference to PUs is limited.
Where a spectrum hole is the band of frequencies assigned to
a PU, but at a particular time and specific geographic location, To effectively implement the concept of CR networking,
the band is not being utilized by that user[5]. Spectrum sensing CR systems need the capability to perform the following
is the first important challenge in CR since all other functions functions :
are on the basis of that. Spectrum Sensing
This paper uses the Spectrum Sensing based on Wavelet
Packet Entropy of the received signal for OFDM signals con- Spectrum Decision
sidering Rayleigh channel model . The method is Blind,ie,no
Spectrum Sharing
need of any prior knowledge of signal or noise estimate
in the channel.Simulation results shows that the method is Spectrum Mobility.
Natonal Conference on Signal & Image Processing -2014 (SIP) 61
In spectrum sensing, CR users sense the PU spectrum occu- spectrum of OFDM signals can be adaptively shaped to fit into
pancy status and recognize the spectrum holes in the licensed the required spectrum mask. Since the location of interference
bands that can be used for their own communications. Based is known spectrum mask can simply applied to subcarriers
on the sensing results, CR users determine which spectrum [2].
band to use (spectrum decision), how to share the spectrum An example of spectrum sensing and shaping procedures in
with other CR users (spectrum sharing), and when to evacuate OFDM-based CR systems is illustrated in Figure 1
the current spectrum band for the returned PUs (spectrum In this Paper, we suggest the extension of Wavelet Packet
mobility).Spectrum Sensing is the most important part in entropy based method for Sensing in ODFM based Cognitive
Cognitive Radio because based on this all other functions are Radio environment.
carried out.
IV. WAVELET PACKET E NTROPY BASED S PECTRUM
III. OFDM BASED C OMMUNICATION IN C OGNITIVE S ENSING
R ADIO
Due to noise uncertainty, the performance of traditional
OFDM is a multi carrier modulation technique that can detectors deteriorates rapidly at low Signal to Noise
overcome many problems that arise with high bit rate commu- Ratios (SNR). Furthermore, without accurate estimation of
nications. The biggest problem of high bit rate communication background noise, there exists an absolute SNR wall below
is time dispersion.This is removed in OFDM using multiple which a detector may fail to be robust, no matter how long
orthogonal sub carriers.So,multi path echoes affect only a the detector can observe the channel[8].
small portion of the neighboring symbols.Remaining inter-
symbol interference (ISI) is removed by extending the OFDM In [1] a Wavelet Packet Entropy(WPE) based PU detection
symbol with a cyclic prefix (CP).Due to these advantages approach is proposed. In the proposed algorithm, the existence
OFDM is the best technique in implementation of Cognitive of the PU is detected when the wavelet packet entropy of
Radio. received signals is smaller than the threshold . By eliminating
the need for accurate estimation of noise, the proposed
A. OFDM based Cognitve Radio: algorithm is robust against noise uncertainty. Moreover,
Sensing and Shaping utilizing the normalized wavelet packet energy vector of
frequency band of node, we can fast estimate the location
1) Sensing: Here, all the cognitive radios are assumed to for spectrum holes of interested frequency band. Here for a
be communicating with OFDM technology as a Secondary given level j , the wavelet packet transform decomposes the
user in the Unused licensed band.When a Primary user enters noisy signal x(n) into 2j sub bands the corresponding wavelet
into that licensed band it should be detected and CR device coefficient sets as:
then do not cause any interference to the Licensed user.Since
OFDM communication is through a number of subcarriers, the dji,m = W P {x(n), j}, n = 1, ...N (1)
Subcarrier location in spectrum which cause interference must
be identified.This is called sensing in OFDM based CR.
Where dji,m denotes the mth coefficient of the ith subband
for the level j, and m=1, ..., N/2j , i=1,...2j The subband
wavelet packet entropy is defined in terms of the relative
wavelet energy of the wavelet coefficients . The energy for
each subband i and level j can be calculated as:
X
Ei j = |di,m j |2 (2)
m
j
where di,m is given by and the total energy of the wavelet
packet coefficients Etotal j is obtained by
j
X 2
X
j j 2
Etotal = |di | = Ei j (3)
i i=1
Fig. 1. Spectrum sensing and shaping using OFDM. And the probability distribution for each level can be defined
as:
2) Shaping: After a CR system scans the spectrum and Ei j
Pij = (4)
identifies active Primary Users, other rental users, and available Etotal j
opportunities, comes the next step is shaping of the spectrum.
Ideally, it is desired to allow cognitive users to freely use Where Pij represents the normalized wavelet energy. As the
available bands in the spectrum. It is desired to have a flexible normalized wavelet energy vector P1j , P2j ,...cover the entire
spectrum mask and control over waveform parameters such as frequency band of signal, of each sub signal denotes the
signal bandwidth, power level, and center frequency. OFDM probability
P j distribution of energy at the subspace. Clearly,
systems can provide such flexibility due to the unique nature P
i i = 1 .
of OFDM signaling. By disabling a set of sub carriers, the
Natonal Conference on Signal & Image Processing -2014 (SIP) 62
VI. C ONCLUSION
R EFERENCES
[1] Zi-Long Jiang & Qin-Yu Zhang & Ye Wang & Xue-Quan Shang
Wavelet Packet Entropy Based Spectrum Sensing in Cognitive Radio
IEEE 3rd International Conference on Communication Software and
Networks (ICCSN),2011
[2] Hisham A. Mahmoud & Tevfik Yucek & Huseyin Arslan,OFDM for
cognitive radio: merits and challenges,IEEE Wireless Communications
Magazine,vol.16,issue.2,pp.6-15,2009
[3] Federal Communications Commission,Spectrum Policy Task Force,
Rep.ET Docket no. 02-135, Nov. 2002.
[4] Boswami, Jaideva C: Fundamentals of Wavelets, theory, algorithms, and
application.
[5] S.Haykin,Cognitive radio:Brain empowered wireless communica-
tions,IEEE Journal on Selected Areas in Communications, vol. 23, no.
2, pp. 201-220, 2005.
[6] J. G. Proakis, Digital Communications, 4th ed. McGraw-Hill, 2001
[7] J. Mitola and G. Q. Maguire Jr., Cognitive radio: Making software radios
more personal IEEE Personal Communications, vol. 6, no. 4,pp. 13 -
18, Aug. 1999
[8] T. Yucek and H. Arslan, A survey of spectrum sensing algorithms for
cognitive radio applications, IEEE Commun. Surveys Tuts., vol. 11, no.
1, Jan.-Mar. 2009, pp. 116-130.
Natonal Conference on Signal & Image Processing -2014 (SIP) 64
I. I NTRODUCTION
Population agingis a serious problem the world is going
to face in near future. The number of people aged 65 plus
will be nearly 1.5 billion in 2050 compared to 524 million in
2010 [1]. This situation results in a high demand of health care
facilities.
Wireless body area networks (WBANs) helps to monitor Fig. 1. The geometry of the proposed antenna.
thr patients health related parameters remotely, continuously,
and in real time. Which will increase the efficiency of health-
care [2]. mainly determined by length(Lp ) and width(Wp )of the an-
The major issue with wired body area network is its incon- tenna. The resonant frequency of the antenna determined by
venience. WBAN is proposed to overcome this issue, which [5].
offer mobility for the user. WBAN consist of several wireless
sensor nodes scattered on the human body communicating with
C
a single body central unit [3]. fr = (1)
2Le r
The Ultra-Wideband (UWB) technology can be used for
WBAN since it has ultra low power consumption. The lower
UWB ranges from 3269.8 MHz to 4742.4 MHz and upper where
UWB ranges from 6240 MHz to 10233.6 MHz [4]. Le = Lp + 2L (2)
Major problem with monople UWB antennas which work w
well in both lower and upper UWB band is high SAR value due L (ef f + 0.3)( hp + 0.264)
= 0.412 w (3)
to the partial ground plane. Therefore in this paper a microstrip h (ef f 0.258)( hp + 0.8)
patch antenna for the lower UWB band is proposed and its
biological effects are studied. r + 1 r 1 12h 1
ef f = + (1 + )2 (4)
2 2 wp
II. A NTENNA D ESIGN
The geometry of the proposed microstrip patch antenna where C is the velocity of light. Here fr =4006.1 MHz [4]
is shown in Fig.1. Antenna is fed with two 50 microsrtip .From this rectangle patch U shape is formed as mentioned in
lines on the FR4 substrate with relative permittivity r = 4.4 [5].The parameters (in mm) of the antenna are optimized and
and height h = 1mm The resonant frequency of the antenna the results are shown in Table I.
Natonal Conference on Signal & Image Processing -2014 (SIP) 65
Ls Ws Lp Wp f
R EFERENCES
[1] WHO, Global Health and Aging,NH Publication no.11-7737, Oct.
2011, pp.1-5.
Fig. 5. Simulated 3D Radiation pattern of the antenna at frequencies 3.5, 4, [2] M. Li, W. Lou, and K. Ren, Data security and privacy in wireless body
and 4.5 GHz. area networks, IEEE Wireless Commun., vol. 17, no. 1, Feb. 2010, pp.51-
58.
[3] B. Latre, B. Braem, I. Moerman, C. Blondia, and P. Demeester, A survey
on wireless body area networks, Wireless Networks Journal, vol. 17,
V. C ONCLUSION no.1, Jan. 2011, pp.1-14.
[4] IEEE Standard for Local and metropolitan area networks: Part 15.6
Microstrip patch antenna which works well in lower UWB Wireless Body Area Networks, IEEE Standard 802.15.6-2012, Feb. 2012.
band is designed. Radiation pattern is omnidirectional, and [5] Atser A. Roy, Joseph M. Mom, Gabriel A. Igwue, Enhancing the
Bandwidth of a Microstrip Patch Antenna using Slots Shaped Patch,
slight variation in the higher frequency. SAR found for two ori- in AJER, vol.2, issue.9 2013, pp.23-30.
entations of the antenna. Parallel placed antenna have less SAR
[6] Omkar Pradhan, Parametric analysis of meandered inverted-F antenna
value, compared to vertically placed antenna. The maximuum and use of a High impedance surface based ground plane for WBAN
value of input power that can be given to the antenna are applications, in Proceedings of IEEE International Conference on Body
found to be 712W and 52W respectively for parallel placed Sensor Networks(BSN), May 2013, pp.1-7.
antenna and vertically placed antenna. These characteristics [7] ifac, Tissue properties, [Online].Available:http//niremf.ifac.cnr.it/tissprop.
Natonal Conference on Signal & Image Processing -2014 (SIP) 67