Signal Processing 4 the Networked Battlespace
Lead Research Organisation:
University of Edinburgh
Department Name: Sch of Engineering
Abstract
Sensors have for a long time played a vital role in battle awareness for all our armed forces, ranging from advanced imaging technologies, such as radar and sonar to acoustic and the electronic surveillance. Sensors are the "eyes and ears" of the military providing tactical information and assisting in the identification and assessment of threats. Integral in achieving these goals is signal processing. Indeed, through modern signal processing we have seen the basic radar transformed into a highly sophisticated sensing system with waveform agility and adaptive beam patterns, capable of high resolution imaging, and the detection and discrimination of multiple moving targets.
Today, the modern defence world aspires to a network of interconnected sensors providing persistent and wide area surveillance of scenes of interest. This requires the collection, dissemination and fusion of data from a range of sensors of widely varying complexity and scale - from satellite imaging to mobile phones. In order to achieve such interconnected sensing, and to avoid the dangers of data overload, it is necessary to re-examine the full signal processing chain from sensor to final decision.
The need to reconcile the use of more computationally demanding algorithms and the potential massive increase in data with fundamental resource limitations, both in terms of computation and bandwidth, provides new mathematical and computational challenges. This has led in recent years to the exploration of a number of new techniques, such as, compressed sensing, adaptive sensor management and distributed processing techniques to minimize the amount of data that is acquired or transmitted through the sensor network while maximizing its relevance. While there have been a number of targeted research programs to explore these new ideas, such as the USs "Integrated Sensing and Processing" program and their "Analog to Information" program, this field is still generally in its infancy.
This project will study the processing of multi-sensor systems in a coherent programme of work, from efficient sampling, through distributed data processing and fusion, to efficient implementations. Underpinning all this work, we will investigate the significant issues with implementing complex algorithms on small, lighter and lower power computing platforms. Exemplar challenges will be used throughout the project covering all major sensing domains - Radar/radio frequency, Sonar/acoustics, and electro-optics/infrared - to demonstrate the performance of the innovations we develop.
Today, the modern defence world aspires to a network of interconnected sensors providing persistent and wide area surveillance of scenes of interest. This requires the collection, dissemination and fusion of data from a range of sensors of widely varying complexity and scale - from satellite imaging to mobile phones. In order to achieve such interconnected sensing, and to avoid the dangers of data overload, it is necessary to re-examine the full signal processing chain from sensor to final decision.
The need to reconcile the use of more computationally demanding algorithms and the potential massive increase in data with fundamental resource limitations, both in terms of computation and bandwidth, provides new mathematical and computational challenges. This has led in recent years to the exploration of a number of new techniques, such as, compressed sensing, adaptive sensor management and distributed processing techniques to minimize the amount of data that is acquired or transmitted through the sensor network while maximizing its relevance. While there have been a number of targeted research programs to explore these new ideas, such as the USs "Integrated Sensing and Processing" program and their "Analog to Information" program, this field is still generally in its infancy.
This project will study the processing of multi-sensor systems in a coherent programme of work, from efficient sampling, through distributed data processing and fusion, to efficient implementations. Underpinning all this work, we will investigate the significant issues with implementing complex algorithms on small, lighter and lower power computing platforms. Exemplar challenges will be used throughout the project covering all major sensing domains - Radar/radio frequency, Sonar/acoustics, and electro-optics/infrared - to demonstrate the performance of the innovations we develop.
Planned Impact
The impact of the proposed research will be manifold. The military have an aspiration of a network of interconnected sensors providing persistent and wide area surveillance of scenes of interest, and to this end have invested heavily in a Network Enabled Capability (NEC) of satellites and ground stations linking to unmanned vehicles, platforms and drones. Such a capability will provide an important operational advantage to our armed forces. The research in this project, examining the acquisition and processing of information from multi-sensor systems in an integrated manner, will help them realize this capability.
Furthermore, as the research project aims to work closely with the UK defence industries there is likely to be a significant economic benefit. A successful project will translate into greater technology pull through of our research ideas into the commercial sector and will help our defence companies remain at the leading edge in the international defence market.
Beyond the defence sector this work will impact on many other disciplines that will benefit from the development of improved sensing systems, with reduced computation and power costs, including: robotics, astronomy, remote sensing and medicine.
Finally our proposed research sits right at the heart of the EPSRC intelligent information infrastructure (TI3) vision to "intelligently manage massive amounts of data, ensure efficient communications and exploit the content and information that will be available." It will therefore impact on a range of related academic areas from mathematics and statistics, through computer science to the engineering of novel sensing systems in areas such as surveillance and medicine.
Furthermore, as the research project aims to work closely with the UK defence industries there is likely to be a significant economic benefit. A successful project will translate into greater technology pull through of our research ideas into the commercial sector and will help our defence companies remain at the leading edge in the international defence market.
Beyond the defence sector this work will impact on many other disciplines that will benefit from the development of improved sensing systems, with reduced computation and power costs, including: robotics, astronomy, remote sensing and medicine.
Finally our proposed research sits right at the heart of the EPSRC intelligent information infrastructure (TI3) vision to "intelligently manage massive amounts of data, ensure efficient communications and exploit the content and information that will be available." It will therefore impact on a range of related academic areas from mathematics and statistics, through computer science to the engineering of novel sensing systems in areas such as surveillance and medicine.
Organisations
Publications
Mehrdad Yaghoobi
(2016)
Range Focusing in Volumetric SAR: a Phase Recovery Approach
Pailhas Y
(2016)
Neither PAS nor CAS: MIMO
Borgia, Alessandro;
(2017)
A Tale of Two Losses: Discriminative Deep Feature Learning for Person Re-Identification.
Narykov A
(2017)
Second-Order Statistics for Threat Assessment with the PHD Filter
Pailhas Y
(2014)
Large MIMO sonar systems: A tool for underwater surveillance
Garcia P
(2017)
Learning to Approximate Computing at Run-time
Voulgaris K
(2021)
DeepMP for Non-Negative Sparse Decomposition
Calum Blair
(2015)
Identifying anomalous objects in SAS imagery using uncertainty
Yaghoobi M
(2016)
Phase recovery for 3D SAR range focusing
Pailhas Y
(2015)
MIMO sonar systems for harbour surveillance
Pailhas Y
(2015)
Orthogonal waveforms for large MIMO sonar systems
Yaghoobi M
(2015)
Fast non-negative orthogonal least squares
Nicolas Valeyrie
(2017)
ONMEX'16 and MANEX'16 MCM trials using UWBMBS (Ultra WideBand MultiBeam Sonar)
Shoukry H
(2017)
Non-Cooperative Target Localisation Using Rank Based EDM Approach
Chenot, C
(2018)
ANOMALY DETECTION WITH HIGH RESOLUTION HYPERSPECTRAL OBSERVATIONS
Pailhas Y
(2015)
Wideband CDMA Waveforms for Large MIMO Sonar Systems
Blair C
(2015)
GPU-Accelerated Gaussian Processes for Object Detection
Rusu C
(2016)
Learning Fast Sparsifying Transforms
Description | This project has now reached completion Research into low complexity sub Nyquist samplnig using efficient fractional delays through delay compensated TF transforms has offered some novel ideas. Our distributed multi-sensor processing work has moved on considerably to provide improved scalability and has overcome issues of resource limitations such as communication bandwidth and power. Current research focus in underwater sonar systems and in particular Multiple Input Multiple Output (MIMO) systems has developed new algorithms which enhance the capabilities and performances of current MIMO systems. Our work on audio-video tracking has focused on the development of new probabilistic behaviourial recognition models that have led to a number of publications. Novel algorithms for multiple target tracking has been developed in a closed-loop sensor management context. Work has also focused on efficient implementations of Gaussian Processes to support classification with confidence. There has already been considerable exploitation and technology transfer from this project - see narrative impact for details. The success of the research outputs from the UDRC project were showcased at an end of project event and has resulted in the award of the 3rd phase of the UDRC research programme |
Exploitation Route | It has resulted in the award of the 3rd phase of the UDRC research programme |
Sectors | Aerospace Defence and Marine Digital/Communication/Information Technologies (including Software) |
URL | http://www.mod-udrc.org |
Description | The research from this project has led to a number of enabling contracts with Dstl, these are listed below: ED TIN 2-1 Tracking and Association- State of the Art Review, £12k + VAT. This is now completed, follow-on work and has led to ED TIN 2-5. This project looked at multi-target tracking algorithms for space surveillance applications and key methods were presented for this scenario. ED TIN 2-2 Application of novel tracking and association methods for Space Situational Awareness (SSA), £25k + VAT. This is now complete. Processed raw data from ground-based radars and optical sensor systems provided by Dstl, and proposed the corresponding realistic sensor models in order to solve the multi-orbit determination problem. This illustrated the novel filtering framework works on real data. ED TIN 2-3 Innovative underwater track motion analysis concepts, Research Associate employed for 1 year, £127k + VAT for 12mths: Project Started 5/1/15 and ended 30/12/15. This project identified an algorithmic solution to multi-sensor and tracking challenges in submarines and this work will be utilised in designing the submarines of the future. ED TIN2-4 Spectral De-convolution, £13k + VAT. Project completed and further work see ED TIN2-6. Designed a raman spectral deconvolution model and tested successfully on Dstl data. ED TIN 2-5 Tracking and Association - State of the Art Review £115057.50, in progress. Follow-on work looking at more specific algorithms, finite set statistics and applying to Dstl data for space surveillance challenges. ED TIN2-6 Deconvolution of Raman Spectral Mixtures 2, £36550 in total, complete. Developed a greedy sparse raman spectral unmixing algorithm with a simple baseline correction step with a low memory usage. Dstl have set up contract with UoE and Snowie Range Instruments to develop demonstrator. ED TIN2-7 Temporal Anomaly Detection challenge,£85787.50, started and due to finish 31/12/16. The aim of this project is to provide signal processing algorithms to discover anomalies in noisy video surveillance data given by Dstl. ED TIN2-8 Mobile Ad-Hoc Sensor Network (MASNET) Modelling Task, £120K, a researcher will be employed for 12 months, started - kick off meeting March 2016. The aim of this work is to the characterise the use of a Mobile Ad Hoc network in two generic scenarios; rural and urban. ED TIN2-9 Chirplet Transform Technique - Further development of a chirplet transform technique for the detection and characterisation of radar signals, £7785 + VAT: Complete. Review and tested code and delivered with the demonstration scripts and provide technical support as required when Dstl integrate with the QinetiQ test bench. |
First Year Of Impact | 2013 |
Sector | Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software),Electronics |
Impact Types | Policy & public services |
Description | udrc annual summer school in defence signal processing |
Geographic Reach | National |
Policy Influence Type | Influenced training of practitioners or researchers |
Impact | Provided training in signal processing for defence aimed at PhD students, industry and government |
URL | https://udrc.eng.ed.ac.uk/ |
Description | Signal Procssing in the Information Age |
Amount | £4,092,207 (GBP) |
Funding ID | EP/S000631/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 06/2018 |
End | 03/2024 |
Description | |
IP Reference | |
Protection | Copyrighted (e.g. software) |
Year Protection Granted | |
Licensed | Yes |
Impact | Licened SAR imaging software to SEA Ltd. for 2014-15 and proviided follow on consultancy on compressed sensing for 3Dlow frequency SAR |
Description | |
IP Reference | |
Protection | Copyrighted (e.g. software) |
Year Protection Granted | |
Licensed | Yes |
Impact | University of Edinburgh has licenced the udrc developed algorithm for separating components of Raman spectral mixtures to Metrohm Raman inc. |
Title | Approximate Adaptive Beam-Pattern Design |
Description | Matlab code supporting the publication: Herbert, Hopgood and Mulgrew, "Computationally simple MMSE (A-optimal)adaptive beam-pattern design for MIMO activesensing systems via a linear-Gaussian approximation", IEEE Transactions on Signal Processing (in submission). |
Type Of Technology | Software |
Year Produced | 2018 |
Description | EURASIP-UDRC summer school in defence signal processing |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | The EURASIP-UDRC summer school is a free to attend summer school in signal processing and AI for defence, training around 75 people per year - a mix of PhD students and early career professionals from the defence community |
Year(s) Of Engagement Activity | 2013,2014,2015,2016,2017,2018,2019,2021,2022,2023 |
URL | https://udrc.eng.ed.ac.uk/events |
Description | Research Themed meetings |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Industry/Business |
Results and Impact | The UDRC run bi-annual research themed meetings for the defence industry and government on topics related to signal processing for defence. We have run UDRC Themed meetings for both grants "Signal Processing in a networked Battlespace" and Signal Processing in the Information age". For the grant Signal Processing in a networked Battlespace, we have delivered the following themed meetings: • Source separation and sparsity • Network and Information Sciences International Technology Alliance • Autonomous systems and signal processing • Hardware and implementation • Image and video processing • MIMO and radar signal processing • Uncertainty and anomaly detection • Space surveillance and tracking • Underwater sensing, signal processing and communications • Data science and signal processing (with Alan Turing Institute) For the grant Signal Processing in the Information Age we have delivered the following themed meetings: 1. Scalable Signal Processing with Bayesian Graphical Models and was held on 20th February 2019. 40 people attended and there was a good mix of academia and defence industry (50:50 split). 2. Deep Learning and Defence was held on 14th November 2019. 70 people attended and there was a good mix of academia and defence industry (50:50 split). 3. UDRC Themed Meeting on Imaging through Obscure Media on 22nd July 2020 - online and 66 attendees (50:50 split). 4. UDRC Themed Meeting on Electromagnetic Environment on 25th November 2020 - - online and 110 attendees (50:50 split). 5. UDRC Themed meeting on Underwater Signal Processing on 25th March 2021 - online and 140 attendees (50:50 split). 6. UDRC Themed meeting on Autonomous Systems on 24 November 2021 - Heriot-Watt University and online and 90 attendees (50:50 split). 7. UDRC Themed meeting on Multiple Object Tracking and Decentralised Processing on 14th January 2022 - online and 90 attendees (50:50 split). 8. UDRC Themed meeting on Algorithm Implementation and Low SWAP Challenges on 30th November 2022 - in person event and 49 attendees (50:50 split) 9. UDRC Themed meeting on Quantum Sensing and Signal Processing on 3rd Mar 2023 - in person event and 36 attendees (70% academia and 30% industry). |
Year(s) Of Engagement Activity | 2013,2014,2015,2016,2017,2018,2019,2020,2021,2022,2023 |
URL | https://udrc.eng.ed.ac.uk/ |
Description | Sensor Signal Processing for Defence Conference (SSPD) |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Industry/Business |
Results and Impact | The SSPD is held annually and organised across the two grants listed. The conference has Industrial and Military Sessions and is tecnically sponsored by the IEEE signal Processing Society . All the presentations and posters can be found on the SSPD website and the papers are now published in IEEE XPlore. The 2019 event was held in Brighton on the 9th and 10th May and we had 122 people attending. Al Hero from the University of Michigan and Andy Bell from Dstl were the keynote speakers. We also had a number of invited Speakers: Daniele Faccio, University of Glasgow; Simon Maskell, University of Liverpool; and, Peter Willet, University of Connecticut. 29 Papers were submitted to the conference and 23 papers were accepted. SSPD2020 was an online conference and attracted 153 attendees. Keynote speakers were: Vivek Goyal, University of Boston; Daniel D. Sternlicht, Naval Surface Warfare Center Panama CityInvites speakers were Paul Thomas, Dstl; Athina Petropulu, Rutgers University; Sean Gong, Queen Mary University of London; Paul White, University of Southampton. SSPD2021 was held as a hybrid conference, Edinburgh and online. René Vidal from Johns Hopkins Mathematical Institute for Data Science and High Griffiths from Defence Science Expert Committee (DSEC) / University College were old keynote speakers. Invited speakers were Alan Hunter, University of Bath; Tien Pham, (CISD) U.S. DEVCOM ARL and Mark Briers, The Alan Turing Institute. There were 138 attendees. SSPD2022 was a hybrid conference, London and online. Lance M. Kaplan, ARL and Frédéric Barbaresco from Thales were our keynote speakers. Simon Godsill, University of Cambridge and Jon Spencer, Dstl Comms & Nets Programme Chief Scientist were are invites speakers. There were 112 attendees. SSPD2023 was a in person conference held in Edinburgh. Prof Jason Ralph from the University of Liverpool and Dr Paul Caseley from Dstl were our keynote speakers and two invited speakers on Machine Learning Techniques for Detecting Hostile Signals, Prof. Kin Leung and Dr Thanos Gkelias from Imperial College London and one invited speaker, Dr Alex Serb from the University of Edinburgh on Adiabatic computing for low power image sensing. There were 83 attendees. |
Year(s) Of Engagement Activity | 2013,2014,2015,2016,2017,2018,2019,2020,2021,2022,2023 |
URL | http://www.sspdconference.org |
Description | UDRC Summer School |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Industry/Business |
Results and Impact | The Summer school is organised under the two grants listed. In previous schools, we have also successfully applied for EURASIP funding. Topics taught: • Statistical Signal Processing • Sensing and Tracking • Machine Learning • Source Separation and Beamforming People who attended the Summer school fed back that they really enjoyed it. A few of their comments are below: • Really enjoyed each day; the lectures were all well suited to my level of experience • High level of content, informing attendees of concepts and state-of-the-art |
Year(s) Of Engagement Activity | 2013,2014,2015,2016,2017,2019,2021,2022 |
URL | https://udrc.eng.ed.ac.uk/events |