University Defence Research Centre (UDRC) in Signal Processing at Imperial College

Processing of real-world analogue signals measured using a variety of sensors, such as audio microphones, and which propagate in multipath or reverberant environments is fundamental to a variety of applications. Within civilian and domestic settings it is important in applications such as for indoor teleconferencing and hands-free audio enhancement. Within the homeland security and defence sectors it is crucial in a wide variety of fields such as forensics and surveillance, both indoors and outdoors, as well as a number of problems requiring target detection and identification: for example outdoor gunshot detection and sniper localisation in urban environments.  

Any signal radiated in a confined space exhibits reverberation, also known as multipath propagation, due to reflections off surrounding obstacles. Despite this being a well understood phenomenon, many existing signal processing technologies fail to explicitly model the multipath response. Blind multipath equalisation can be improved with accurate channel modelling which, in turn, depends on knowledge of the target-position, thereby requiring target tracking. However, many passive target tracking methods suffer from the presence of multipath leading to substantial tracking errors. Target tracking can thus be improved by modelling the effect of, or even equalising, the acoustic reverberation from the observations, thereby allowing identification of the true source from signal reflections. Target tracking and blind multipath equalisation should therefore be solved jointly rather than separately. 

The objective of this 12-month research programme is to address the detrimental effect of multipath mitigation by developing algorithms for joint blind channel estimation and passive source localisation of speech sources in an indoor multipath environment. This proposal therefore covers the general UDRC research requirement topics of broadband signal separation, detection, high-resolution localisation, multipath mitigation, and nonstationary processing. In addition to domestic and homeland security applications, the proposed research will lead to solutions and insight for addressing problems arising in underwater acoustics such as propagation induced distortion in Challenge 16, as well as the joint detection and tracking problems of Challenge 14.

Project Supervisor

Dr. James Hopgood

James Hopgood is a lecturer in the Institute for Digital Communications, within the School of Engineering, at the University of Edinburgh, Scotland. His research interests include nonstationary signal processing, speech and audio signal processing in adverse acoustic environments including blind reverberation and acoustic source localisation, single channel signal separation, medical imaging, and general statistical image processing. James received the M.A., M.Eng. degree in Electrical and Information Sciences in 1997 and a Ph.D. in July 2001 in Statistical Signal Processing, part of Information Engineering, both from the University of Cambridge, England. He was then a Post-Doctoral Research Associate for the year after his Ph.D within the same group, at which point he became a Research Fellow at Queens' College continuing his research in the Signal Processing Laboratory in Cambridge. James joined the University of Edinburgh in April 2004.

The goal of this project is to develop a general detection theory to guide the processing of images from synthetic aperture radar (SAR) along track interferometers (ATI) with greater than two beams.  Such a detection theory should allow the construction of constant false alarm rate (CFAR) detection rules for small, slow moving targets in high clutter environments.  The technical approach taken is to decompose the multi-beam covariance matrix via an Eigen decomposition and investigate the use of the decomposition parameters and Eigenvalues as ground moving target indication (GMTI) metrics.

Project Supervisor

Dr. Jonathan Barker

Jon Barker is a member of the Networked Sensors and Fusion Team in the Sensors and Countermeasures Department at the Defence Science and Technology Laboratory. He received his Ph.D. in stable homotopy theory in 2006 from the University of Southampton. His current research interests lie in multi-beam SAR processing for GMTI; Bayesian information fusion; and, statistical anomaly detection.

Radar systems placed in the nose of fast moving jets have to detect moving targets, the radial velocity of which is close to that of the surrounding clutter, relative to the platform's speed. A widely used moving target detection strategy for moving platforms is that of space-time adaptive processing (STAP). STAP is now readily applied to the case of a sideways-looking array, where the majority of the clutter occurs along a narrow ridge, which crosses the angle-Doppler graph diagonally. When the array orientation is at an angle (termed the crab angle) to the direction of platform motion, the clutter at a given range no longer occupies the diagonal ridge, but an ellipse. The eccentricity of the ellipse decreases as the crab angle increases, so that when the array is forward-facing with respect to the platform motion, clutter forms a circle on the angle-Doppler plot. It is far more difficult for the STAP to compensate for this clutter because it is now range dependant. As such, ground clutter is range ambiguou, and the clutter arcs at different ranges and angles can interfere with the detection and tracking of targets. Thus the performance of the radar is reduced because of the increased clutter power competing with the target's signal. Current research has concentrated on altering the STAP architecture to cope with the range dependent returns. However, there is already a mechanism which can help mitigate the Doppler-range ambiguities, but which is not used in the adaptive part of the STAP architecture. This, of course, is the matched filter and its ambiguity function. This research will develop methods to efficiently and adaptively design the transmitted waveform based on the received signals. The study will encompass the use of the received signals, prior knowledge of target and clutter locations, and spatial beam pattern of the array on transmit and receive, in designing the signal.

Project Supervisor

Dr. Mathini Sellathurai

Dr. Mathini Sellathurai is currently a Reader in Digital Communications and Signal Processing with the Institute of Electronics, Communications, and Information Technology, School of Electronics, Electrical Engineering, and Computer Science, Queen’s University Belfast, Belfast, U.K. Her current research interests include adaptive and statistical signal processing, space-time and MIMO communications theory, information theory and cognitive radio with applications in radar technology, telemedicine, underwater communications, satellite communications, and future wireless networks. Her current research has been funded by EPSRC under EP/D07827X/1 for advanced signal processing techniques for multi-user multiple-input multiple-output broadband wireless communications; MIMO-RADAR and waveform agility to improve low elevation target detection (also partially supported by QinetiQ, Portsmouth); EP/G026092/1 for Bridging the gap between design and implementation of soft-detectors for Turbo-MIMO wireless systems and EP/H012257/1 for Signal Processing Techniques to Reduce the Clutter Competition in Forward Looking Radar. She is also receiving European Union Frame Work 7 Funding for Cognitive radio oriented wireless networks. Dr. Sellathurai was the recipient of the Natural Sciences and Engineering Research Council of Canada’s doctoral award for her Ph.D. dissertation and a co-recipient of the IEEE Communication Society 2005 Fred W. Ellersick Best Paper Award. Dr. Sellathurai is currently serving as an Associate Editor for the IEEE TRANSACTIONS ON SIGNAL PROCESSING and also an organizer for the IEEE International Workshop on Cognitive Wireless Systems, IIT Delhi, India, 2009. She has been a Technical Program Committee member for the IEEE International Conference of Communications from 2004 to 2010.

Source separation is a critical early processing stage in electronic surveillance systems where the multiple simultaneously intercepted transmissions need to be detected, separated and identified for possible threats (e.g. pulsed and continuous wave radar, navigation systems, etc.). When the signals to be detected and separated overlap in time and frequency this can prove a challenging signal processing task that cannot be solved through simple filtering or beamforming. Recently sparse representations have emerged as a very powerful technique for solving source separation problems, particularly in underdetermined scenarios (i.e when there are fewer target sources than sensors), including the difficult case of single channel source separation. Sparse representations usually exploit prior knowledge of the nature of the signals to be intercepted to create 'nonlinear' separation algorithms that substantially surpass the performance of traditional filtering techniques. Furthermore, in certain circumstances, they can also be adapted to learn the structure of the signals being observed to achieve the separation in a totally blind manner.

The aim of this project is to develop new algorithms based around sparse representations capable of detection, separation and classification of individual EM signals that overlap in time and frequency. In addition computational efficiency will be pursued by borrowing recent ideas from compressed sensing theory.

Project Supervisor

Prof. Michael Davies

Mike Davies received the B.A. (Hons.) degree in Engineering from Cambridge University, Cambridge, U.K., in 1989 and the Ph.D. degree in nonlinear dynamics and signal processing from University College London, London (UCL), U.K., in 1993. Mike Davies was awarded a Royal Society Research Fellowship in 1993. He currently holds the Jeffrey Collins SFC funded chair in Signal and Image Processing at the University of Edinburgh and is the Director of the Joint Institute in Signal and Image Processing, part of the Edinburgh Research Partnership. He is currently pursuing a programme of research in the application of sparse representations to signal processing. Most recently his research has concentrated on the emerging field of compressed sensing.

Synthetic Aperture Radar (SAR) provides the military with an extremely valuable means of remote imaging and plays an important role in target detection. SAR works by measuring the electromagnetic signal reflections from the ground. Processing the raw received data to generate the image is usually performed using linear frequency domain techniqes such as the Polar Format Algorithm. However when there is missing data, or when gaps in the data (either spatially or spectrally) are introduced the performance of such estimators deteriorates considerably. The resulting images exhibit distortion and aliasing artefacts.

Recently a new theory for signal reconstruction, called compressed sensing, has emerged. It explores the extent to which ill-posed sampling problems such as those discussed above can be made well-posed through the inclusion of strong signal models. These techniques have already been successfully applied to SAR image reconstruction for target detection and super resolution. The broad aim of this proposal is to explore the application of compressed sensing reconstruction techniques to SAR image formation when spatial and/or frequency notches are introduced into the transmitted/received signals. Ultimately we hope to gain a general understanding of the limits to which the SAR data acquisition system can be so modified without incurring serious performance degradation.

Project Supervisor

Prof. Michael Davies

Mike Davies received the B.A. (Hons.) degree in Engineering from Cambridge University, Cambridge, U.K., in 1989 and the Ph.D. degree in nonlinear dynamics and signal processing from University College London, London (UCL), U.K., in 1993. Mike Davies was awarded a Royal Society Research Fellowship in 1993. He currently holds the Jeffrey Collins SFC funded chair in Signal and Image Processing at the University of Edinburgh and is the Director of the Joint Institute in Signal and Image Processing, part of the Edinburgh Research Partnership. He is currently pursuing a programme of research in the application of sparse representations to signal processing. Most recently his research has concentrated on the emerging field of compressed sensing.

The goal of this project is to develop novel low-complexity beamforming algorithms based on low-rank decompositions and the set-membership filtering (SMF) framework in order to design adaptive beamformers with complexity one order of magnitude lower than existing techniques. The proposed low-rank decompositions will be based on iterative switching and pattern matching and approximation of basis functions, and do not require complex eigen-decompositions or expensive operations. These techniques can be significantly simpler than full-rank filtering algorithms by reducing the dimensionality of the input data vector. The SMF concept will then be used to design low-complexity adaptive algorithms for the updates of the transformation matrix that performs dimensionality reduction and the low-rank filter. We will formulate the LCMV beamforming problem with the low-rank decompositions using linear algebra, develop SMF-based adaptive algorithms and build simulation tools to design, test and analyse the proposed techniques. The outcomes will be better, simpler and practical beamforming algorithms, and high-quality publications.

Project Supervisor

Dr. Rodrigo C. de Lamare

Rodrigo C. de Lamare was born in Rio de Janeiro, Brazil, in 1975. He received his Diploma in Electronic Engineering from the School of Engineering of the Federal University of Rio de Janeiro (UFRJ) in 1998 and the MSc and PhD degrees in Electrical Engineering from the Pontifical Catholic University of Rio de Janeiro (PUC-RIO) in 2001 and 2004, respectively. He then worked as a Postdoctoral Fellow from January to June 2005 at the Centre for Telecommunications Studies (CETUC), PUC-RIO and from July 2005 to January 2006 at the Signal Processing Laboratory, UFRJ. In 2006, he been a visiting Professor at the University of Oslo, Norway. Since January 2006, he has been with the Communications Group, Department of Electronics, University of York, where he is currently Lecturer in Communications Engineering. His research interests lie in communications and signal processing, areas in which he has published over 150 papers in international journals and conferences.

The aim of this proposed work is to develop efficient generic modules and architectures for integration of the branches of an array-system, as well as on proposing novel concepts to further reduce the size, weight and the power consumption of array-sensor system for target localization and beamforming.
Beamformers have been widely used in military and commercial applications due to their many capabilities in obtaining information from the waveforms received at the array elements in a signal environment which consists of multiple emitting sources (targets) plus noise. In general this is achieved by using an array-pattern-forming-network which places relatively high gain in those directions which contain the desired signal and, at the same time, places nulls in the directions of the interferences. This will increase the detection gain of the “desired” target in a multitarget environment.

However, one main problem with conventional  beamformers (time-domain beamformers) is related to their inability to detect and resolve sources/targets located close together in space. This inability is imposed from the fact that their performance is limited by the Signal-to-Noise-plus-Interference Ratio (SNIR). This gave rise to a new class of techniques, known as High-Resolution or Superresolution spatiotemporal techniques, which are mainly used in the Direction Finding Problem and beamforming.

Project Supervisor

Prof. Eric Yeatman

Eric M. Yeatman has been a member of academic staff in Imperial College London since 1989, and Professor of Micro-Engineering since 2005. He is deputy head of the Department of Electrical and Electronic Engineering, and has published more than 130 papers on optical devices and materials, and micro-electro-mechanical systems (MEMS). He holds several patents, and is co-founder and chairman of Microsaic Systems Ltd., a MEMS product development company spun-out of the college. He has been principal or co-investigator on more than 20 research council and industry supported projects, with over £6M in research funds raised. He has acted as a design consultant for several international companies, and technical advisory board member to two venture capital funds. His current research interests are in radio frequency and photonic MEMS devices, and energy scavenging for wireless sensor nodes.

This project is concerned with the multi-target estimation using the MIMO radar system with co-located arrays in both the homogeneous clutter and non-homogeneous clutter environments where superresolution array processing techniques will be comprehensively investigated. This objective is motivated by the commercial and military needs for the radar systems which have high resolution location estimation and robustness against the jammers and noise. The consequences of the study on such problem will not only provide a basis for improved utilization, but would also lead to a faster expansion and deeper penetration of their applications in both military and civilian aviation.

MIMO radar is an emerging technology that is attracting the attention of researchers and practitioners, which employ multiple transmit signals and have the ability to jointly processing signals received at multiple receive antennas. "Co-located arrays" is a typical con…guration in MIMO radar, which includes a transmitter-array (Tx-array) and a receiver-array (Rx-array) that both have arbitrary geometry. Moreover, the directions of a target with respect to both the Tx-array and the Rx-array are the same.

Project Supervisor

Prof. Athanassios Manikas

Professor Manikas holds the Chair of the Communications and Array Processing in the Department of Electrical & Electronic Engineering, Imperial College London. He has published an extensive set of journal and conference papers in the area of digital communications and array signal processing and is the Author of a book (monograph) entitled "Differential Geometry in Array Processing". He is on the editorial board of IET Proceedings Signal Processing and the editor of the ICP research book-series on Communications and Signal Processing (jointly with Prof. A.G. Constantinides). He has held a number of research consultancies for the EU, industry and government organisations. He also has had various technical chairs at international conferences and has been a TPC member of major IEEE conferences He has served as an Expert Witness in the High Court of Justice (UK) and is currently a member of the Royal Society's International Fellowship Committee (1st Jan 2008 until 31st Dec 2010). He is leading a strong group of researchers at Imperial College and has supervised successfully 29 PhDs and more than 100 Masters project-students. Professor Manikas is a Senior Member of IEEE, a Fellow of IEE and a Chartered Engineer.

General Aim: This project is concerned with signal processing techniques for on-line calibration and recalibration of sensor arrays in order to compensate errors/uncertainties in location (geometrical errors), phase and gain (electrical errors), synchronisation, etc, even when all errors/uncertainties are present simultaneously and can change with time.

Array of sensors: An "uncalibrated" array of N sensors of arbitrary geometry will be considered operating in a multitarget environment where the array uncertainties/errors can degrade signi…cantly the array performance either slowly or even abruptly. Note that the array input vector-signal may contain also sperious signals due to sensor/hardware problems while sensor failure (partial or full), will be studied as part of the array calibration problem.

In summary, in this project the problem of uncertainties/errors in sensor-arrays will be addressed using "hybrid" auto-calibration approaches. This is a mixture of novel pilot and self calibration techniquescombined in an smart and automated way. The proposed algorithms will be veri…ed using Imperial College hardware platform.

Project Supervisor

Prof. Athanassios Manikas

Professor Manikas holds the Chair of the Communications and Array Processing in the Department of Electrical & Electronic Engineering, Imperial College London. He has published an extensive set of journal and conference papers in the area of digital communications and array signal processing and is the Author of a book (monograph) entitled "Differential Geometry in Array Processing". He is on the editorial board of IET Proceedings Signal Processing and the editor of the ICP research book-series on Communications and Signal Processing (jointly with Prof. A.G. Constantinides). He has held a number of research consultancies for the EU, industry and government organisations. He also has had various technical chairs at international conferences and has been a TPC member of major IEEE conferences He has served as an Expert Witness in the High Court of Justice (UK) and is currently a member of the Royal Society's International Fellowship Committee (1st Jan 2008 until 31st Dec 2010). He is leading a strong group of researchers at Imperial College and has supervised successfully 29 PhDs and more than 100 Masters project-students. Professor Manikas is a Senior Member of IEEE, a Fellow of IEE and a Chartered Engineer.