OFEME 2024 Programme

Richard has worked in Defence research for over 30 years, with experience in all the military domains.  Richard’s career spans a wide range of activities such as strategic decision support, procurement advice, trials, international research collaboration, and operational support, including deployments to Bosnia, Iraq and Afghanistan. For over a decade Richard has worked in programme management and is now one of Dstl’s most experienced programme managers, having delivered a number of different research programmes.  Richard is currently managing MOD’s Communications and Networks research programme, the scope of which includes systems and technologies for wireless communications waveforms and bearers, resilient deployed networks and information services.

Douglas Paul has an MA degree in Physics and Theoretical Physics and a PhD from the Cavendish Laboratory, University of Cambridge. He is Professor of Semiconductor Devices and presently holds a Royal Academy of Engineering Research Chair in Emerging Technologies at the University of Glasgow. He previously held an EPSRC Quantum Technology Fellowship awarded to provide leadership for the UK Quantum Technology Programme and an EPSRC Advanced Research Fellowship at the University of Cambridge. He was the first Director of the James Watt Nanofabrication Centre at the University of Glasgow.

Doug is a Fellow of the Royal Society of Edinburgh, Fellow of the Institute of Physics, a chartered physicist, a chartered engineer and a Senior Member of the IEEE. He was the recipient of the Institute of Physics President’s Medal in 2014. Doug’s research uses micro-fabrication and nano-fabrication approaches to produce practical quantum technology systems for applications including position, navigation, timing, LiDAR, quantum imaging, memories, gravity imaging, sensing and secure communications. He is the principal investigator for the EPSRC Programme Grant “Chip-scale Atomic Systems for a Quantum Navigator” and the UK Hub for Quantum Enabled Position Navigation and Timing which starts in December 2024. He has been principal investigator on over £71M of collaborative grants which included 10 Innovate UK grants with UK industry over the past 5 years and has multiple research contracts from industry.

He presently sits on a number of government advisory committees including the National Security PNT Working Group and previously sat on MOD Defence Science Expert Committee, MOD Defence Science Advisory Council, NATO CBP Science Panel and the Home Office CBRN Scientific Advisory Committee.

Victoria is head of the Defence AI Research Centre (DARe) at the Alan Turing Institute. She oversees research themes in the Electromagnetic Environment, edge AI, Synthetic environments, and AI vulnerabilities and deception. In 2020 she founded the Turing machine learning for radio frequency applications interest group. She focuses her research on applied defence problems and specialises in Synthetic Aperture Radar. Before the Turing Victoria worked at DSTL as a SAR scientist developing novel techniques for geospatial analysis and SAR capability development, and as a data science researcher. Victoria holds a DPhil from the University of Oxford in Earth Sciences specialising in studying intercontinental deformation using InSAR and a degree in geophysics from the University of Edinburgh. Prior to her PhD she worked in the oil and gas industry as a research geophysicist at PGS on the production and development of algorithms for the processing and inversion of the first towed electromagnetic streamer. Her research interests include machine learning, co-design of hardware and software, signal detection and identification, and the synthetic modelling of complex physical processes.

AI is rapidly transforming geospatial analysis and presenting a new paradigm for the adversarial threat landscape. Ensuring wider adoption of AI for detection and classification of entities requires a good understanding of the vulnerabilities associated with its use, its susceptibility to deception and the development of explainable, interpretable, and trustworthy techniques. We explore the ways in which machine learning models increase the attack surface-area of existing systems. This presentation covers adversarial attacks along the entire AI system pipeline, exploiting vulnerabilities in the training phase (e.g. poisoning attacks) and in the testing/ inference phase (e.g. evasion, model extraction and inversion attacks). We focus on the use case of automated approaches for maritime vessel detection and land vehicle classification in synthetic aperture radar (SAR) images. SAR is an all-weather imaging capability that can penetrate cloud cover and be used during the night. Many attack approaches designed for more traditional imaging modalities from open-source libraries e.g. SecML (Melis et al, 2019) can be adapted to SAR data. However, more realistic attack vectors for SAR imaging would require approaches designed for the specific application area e.g. electronic counter measures approaches. We investigate degradation in the performance of machine learning approaches for vessel detection in the xview3 dataset and vehicle classification in the MSTAR dataset. We use a range of techniques including simulation of transponder-based target insertion into SAR collections using the Time Domain Frequency Shift Template Segmentation (TDFS-TS) algorithm (Yang et al, 2019). We then discuss approaches for mitigating against SAR machine learning vulnerabilities and deception.

Dr Gaetano Di Caterina is currently Senior Lecturer, and the Leonardo Lecturer, in the Electronic and Electrical Engineering Department at the University of Strathclyde, where he is Director of the Neuromorphic Sensor Signal Processing (NSSP) Lab, within the Centre for Signal and Image Processing (CeSIP) group. Dr Di Caterina has a background in signal, image and video processing, he is Course Director of the MSc in Machine Learning and Deep Learning at Strathclyde, and has published works on Machine Learning, Deep Learning, Neuromorphic technologies and Signal Processing at several international conferences and in academic journals. Recent research projects include funding from Leonardo UK, DASA/DSTL, AFOSR/AFRL, and ESA.

James R. Hopgood is a Professor of Statistical Signal Processing at the University of Edinburgh (UoE, Scotland) and Director/PI of the EPSRC and MoD Centre for Doctoral Training in Sensing, Processing, and AI for Defence and Security (SPADS). James is also Dean of Quality and Enhancement in the College of Science and Engineering in UoE. Between 2019 and 2024, James was Director of Electronic and Electrical Engineering in the School of Engineering at the UoE, and also a member of the Institute for Imaging, Data, and Communications (IDCOM).
James’s research spans applications of statistical signal processing and machine learning in a diverse range of applications including multi-modal multi-sensor-fusion and multi-target tracking. James has worked with multiple companies translating underpinning statistical signal processing methods to applications such as improved electrophoresis analysis, data driven tracking solutions for magnetometer sensing technology, model-based super resolution imaging techniques for multibeam sonar systems, and developing alternative future concepts of Early Warning systems through the “DASA: Look Out! Maritime Early Warning Innovations”.

The UK government’s fundamental functions of defence and security emphasise the need for information superiority amidst global competition and the AI arms race. Theis is recognised in the Science and Technology Strategy, Defence AI Strategy, and Integrated Review, which stress the importance of AI and data-centric solutions. Information-based defence and security applications rely on interconnected technologies, such as multiple sensing modalities, fusion architectures, processing and communication across algorithms/hardware, autonomy, embedded systems, AI, and beyond CMOS electronics. These technologies share common foundations and require multidisciplinary expertise.

To meet these demands, the EPSRC/MoD Centre for Doctoral Training in Sensing, Processing, and AI for Defence and Security (SPADS), aimed at recent graduates, will boost core engineering and defence-related complementary skills, pursue research with the defence community, and ensure a pipeline of defence-related PhD-level specialists

This talk introduces SPADS, led by the Universities of Edinburgh and Heriot-Watt. SPADS will develop almost 80 PhD graduates, training them to integrate into the UK’s defence ecosystem. Key objectives include funding industry-led research, creating a talent pool, building a network of skilled researchers, and laying foundations for long-term training. SPADS projects will enhance UK technological capability and have broader applications in civilian sectors like healthcare, automotive, and gaming.

Richard is an applied physicist with a focus in working with defence to understand and identify how new technologies can be used to bridge capability gaps, and how to then design a system around them.

Some example of Richard’s recent work include the use of novel computational methods in optimisation and decision support, the design and build of a quantum-enhanced RF sensor, use case and scientific feasibility assessments of novel non-linear optical systems, and identifying how mathematics, physics, photonics and biomimicry can be leveraged to create new technology concepts that work fundamentally differently to current approaches – opening up new trade spaces and new use cases.

Krishna Balram is currently a Professor of Nanoscale Device Engineering at the University of Bristol. His research group works on the development of novel nanofabricated device platforms for manipulating light and GHz sound waves at the nanoscale, and engineering controlled interactions between them and other solid-state quantum systems. He received his PhD from Stanford University, working with David A.B. Miller, and was a postdoc at NIST with Kartik Srinivasan before joining Bristol. His research is supported by an ERC starting and consolidator grant awards. More details can be found at: https://krishnacbalram.github.io/index.html

50+ years of Moore’s law and related advances in nanofabrication have taught us how to make nanoscale objects routinely. But getting information (both classical and quantum) efficiently into and out of these (deeply sub-wavelength) nanoscale objects remains an unsolved problem in general. In this talk, we will discuss one instance of this ‘nanoscale wiring’ problem: interfacing GHz microwave signals efficiently with nanoscale devices using acoustic waves as an intermediary. We show how the same efficiency challenges occur in three very different contexts:
(a) building piezoelectric quantum microwave to optical signal transducers [1,2] for connecting remote superconducting qubit based quantum processors
(b) phononic integrated circuits for re-thinking RF front-ends [3], bringing ideas from silicon photonics to microwave devices
(c) piezoelectric micro-resonators for electron spin resonance (ESR) experiments on single nanoparticles [4], and the prospects for electrical readout of single spins at cryogenic temperatures

References:
[1] K.C. Balram and K. Srinivasan, K. Advanced Quantum Technologies, 5(3), p.2100095, 2002
[2] A. Khurana, P. Jiang, P. and K.C. Balram, Physical Review Applied, 18(5), p.054030, 2022
[3] M. Bicer and K.C. Balram, IEEE Trans. UFFC, 71(1), 172, 2023
[4] C.S. Kline, J.M-Ruz, and K.C. Balram, arXiv:2405.02212, accepted in Phys. Rev. Applied