Communication and signal processing

Non-parametric Smart Sensing Analytics based on Large Spectrum Data and Estimation of Channel Activity Statistics

Future wireless networks will demand huge amounts of radio frequency spectrum resources. It is unlikely that such demand will be met without employing smart dynamic spectrum sharing approaches based on cognitive radio (CR) techniques. In this context, one important requirement of future wireless networks will be the ability to detect the presence of other wireless systems within a particular region of the spectrum occupancy data. Spectrum sensing is a popular approach to address this problem and constitutes a fundamental building block of CR systems. Existing sensing schemes are parametric and imperfect in nature, and unrealistic to implement on large scale networks due to various practical performance limitations. The main objective of this project is to develop feasible non-parametric smart sensing mechanisms with an improved performance obtained by exploiting statistical knowledge of the spectrum activity patterns, and validate their suitability by means of a proof-of-concept wireless prototype / test bed.

This project is funded by DST-UK-India Education and Research Initiative (UKIERI), British Council.

Dr. Dhaval Patel, Dr. Miguel López-Benítez (University of Liverpool, UK), Brijesh Soni (JRF)


Design and Performance Analysis of Non-parametric Detection algorithm for Cognitive Radio - MIMO Communications

The aim of the project is to develop non-parametric detection algorithm and analyze the performance in a real-time wireless environment. In this project, we focus on the study and Monte-Carlo simulations of a new non-parametric scheme to detect the presence of the primary user without knowing its structure and channel information. The work includes the detailed performance evaluation of non-parametric detection scheme in the actual wireless environment. To achieve this, GNURADIO software and the Universal Software Radio Peripheral (USRP) hardware based wireless test bed will be created to prepare the experimental setup of CR-MIMO (Cognitive Radio – Multiple Input Multiple Output) systems. Thus, the project will provide simulations and an experimental performance evaluation of non-parametric detection schemes to provide opportunistic spectrum access in a cognitive radio environment.

This project is funded by Gujarat Council on Science and Technology (GUJCOST), Department of Science and Technology (DST), Government of Gujarat.

Dr. Dhaval Patel, Dr. Sanjay Chaudhary, Mitul Panchal (M.Tech student), Maunil Joshi (B.Tech student)

Learn more


Description based person identification in unconstrained surveillance video

The goal of the project is to locate a person-of-interest based on specific set of soft biometric attributes from a surveillance video without prior registration. The project will integrate and minimize semantic gap between human descriptions and soft biometric traits. Moreover it aims at extracting soft biometric features from an input image or a video frame and then uses these features to locate a matching individual in the video stream. This project will add another dimension to surveillance as human cognitive perceptions are used while searching the video streams.

This project is funded by Board of Research in Nuclear Science.

Dr. Mehul Raval, Dr. Sanjay Chaudhary, Anand Laddha, Shvetal Pandya (SRF)


Securing biometric data using data hiding techniques

Biometrics based authentication systems have been gaining widespread acceptance in the information security domain. However there are several loopholes in biometric systems. This project aims at improving the security and authentication mechanisms of biometric systems by using complementary data hiding techniques. Fragile watermarking will be used to check the integrity and authenticity of biometric templates before identification. This will improve the overall security. 

This project is funded by Board of Research in Nuclear Science.

Dr. Mehul Raval, Priti P Rege, S K Parulekar, Vaibhav B Joshi (SRF)


Disease Detection and Severity Estimation in Cotton Plant from Unconstrained Images

The primary focus of this work is to detect disease and  estimate  its  stage  for  a  cotton  plant  using  images.  Most disease symptoms are reflected on the cotton leaf. Unlike earlier approaches,  the  novelty  of  the  proposal  lies  in  processing  images captured under uncontrolled conditions in the field using normal  or  a  mobile  phone  camera  by  an  untrained  person. Such  field  images  have  a  cluttered  background  making  leaf segmentation  very  challenging.  Using  local  statistical  features first classifier segments leaf from the background. Then using hue and luminance from HSV colour space as features another classifier  is  trained  to  detect  disease  and  find  its  stage.  The developed  algorithm  is  a  generalized  as  it  can  be  applied  for any  disease.  However  as  a  showcase,  in  this  paper  we  detectGrey Mildew; a commonly occurring bacterial disease in NorthGujarat, India.

Dr. Mehul Raval, Dr. Sanjay Chaudhary, Aditya Parikh (B.Tech student)


Kahinee - An Interactive Voice Response System for Rural Healthcare

Kahinee is an Interactive Voice Response(IVR) system  focused  at  increasing  health  awareness  in  rural  citizens of  India.  It  consists  of  two  modules  i.e.  content  delivery  module(CDM)  and  question  answer  module(QAM).  CDM  delivers creatively  designed audio  plays on  various  health  topics  to  the rural citizen. QAM helps rural citizen to ask queries pertaining to health topics and get them answered by doctors in the panel.The  system  is  designed  and  developed  using  the  principles  of User   Centered   Design(UCD),   Information   &   CommunicationTechnology(ICT)  and  Business  Analytics(BA).  The  CDM  was tested  in  a  joint  pilot  with  Barakat  Bundle,  Indian  Institute of  Public  Health(IIPH)  and  Society  for  Education  Welfare  andAction(SEWA)  Rural  in  Bharuch  district,  Gujarat,  India.  This work discuss  system  design,  framework  development,  field  implementation  and  results.  The response  by  rural  citizens  during trial  run  yielded  a  confidence  in  system  and  gave  fresh  insights for further development.

Dr. Mehul Raval, Anmol Anubhai (B. Tech student), Rahul Patel (B. Tech student), Shashwat Sanghavi (B. Tech student), Karima Ladhalani, Barakat Bundle

CheckIt - A low cost mobile OMR system

This paper describesCheckIt: a mobile phone based optical mark recognition(OMR) system which is used for automatic checking of the user response sheets. It exploits prior information about the OMR sheet layout, which helps in achieving high speed and accuracy. The system incorporates following interdependent modules: (i) computer vision and image processing; (ii) computer communication and networking; (iii)database and (iv) user interface. The back end is developed in Python and OpenCV library while the front end is made using HTML and Android. The overall system cost is low as;1. software is developed using open-source technology; 2. it does not necessitate scanning hardware. The desirability and viability aspect of the system development is done based on extensive market survey and after interviewing several stake holders.

Dr. Mehul Raval, Shashwat Sanghavi (B. Tech student), Rahul Patel (B. Tech student), Dhruv Gupta (Teaching Associate)

Your browser is out-of-date!

For a richer surfing experience on our website, please update your browser.Update my browser now!