CLASSIFICATION & MULTI-MODAL PROCESSING THEME
This theme is concerned with the developing new multi-sensor signal processing algorithms for classification of objects/targets using one or more types of signals/information. The objects/targets may be rigid (e.g. a submarine) or diffused (e.g. convoys of vehicles, crowds of people, dust clouds) and a great deal of information can potentially be gained by analysing, for instance, radar or sonar imaging, visual and/or audio signals, micro-Doppler returns from a target which has been illuminated by electromagnetic energy or acoustic wavelength radiation - which this theme aims to exploit.
Some application areas include: Radar, Sonar, Person tracking, Robot Audition.
- [C5] Real-Time Multi-Modal Person Tracking
- [I2] Early Auditory-Visual Integration for Transient Detection
- [I6] Wave Wakes
- [O01] Locally Invariant Signal Processing to Discriminate Between Key Man-made and Natural Features
- [O03] Hamiltonian-based data clustering and classification
- [O04] Advanced High Resolution Methods for Radar Imaging and Micro-Doppler Signature Extraction
- [O11] Multimodal Blind Source Separation for Robot Audition
- [O16] Target Detection in Clutter for Sonar Imagery
- [O17] Target Classification And Tracking Using Acoustic Micro-Doppler Signatures
DISTRIBUTED SIGNAL PROCESSING THEME
This theme is concerned with the investigation, analysis and design of distributed signal processing algorithms for detecting and locating single or multiple targets/sources using a wireless sensor network with constrained resources (e.g. minimum power, cooling requirements and cost of communication between sensors). Amongst other approaches, this involves the development of cooperative, temporal and spatial adaptive distributed and fusion algorithms.
Some application areas include: Persistent surveillance of complex environments using distributed signal processing techniques.
- [O02] Generic Distributed Target Tracking Algorithms in Sensor Networks
- [O05] Cooperative Localisation: Distributed Optimisation with Hypothesis Testing
- [O08] Distributed Signal Processing for Distributed Sensor Networks
- [O14] Distributed and Iterative Processing for Wireless Sensor Networks with Multiple Local Fusion Centres
- [O15] Scalable Information Fusion: Adaptivity for Complex Environments and Secure Data
NON-CONVENTIONAL SIGNALS THEME
This theme is concerned with multi-sensor signal processing algorithms that are capable of operating in the presence of non-"standard" signals/noise. Such signals may be non-stationary, non-Gaussian and/or non-circular, spatially and temporarily distributed with various degrees of correlation. Detection, reception, signal extraction and modelling/simulation of such non-conventional signals is one of the main aims of this theme.
- [C3] Widely Linear Adaptive Processing of Noncircular Complex Signals
- [I3] Extracting tones which vary in frequency from non Gaussian noise
- [I5] Synthetic Noise
- [O12] Real Time Model Adaptation for Non-Stationary Systems
SUPER-RESOLUTION SOURCE SEPARATION THEME
This theme is focusing on super-resolution detection and localisation/separation of sources (emitters/targets/signals) located close together in space (angle and/or range) using a multi-sensor system.
The resolution performance of such a system is, in general, a function of its aperture and number of sensors, N. For instance, the larger the multi-sensor aperture the better the resolution performance. In practice, these resources are limited and so the purpose of a super-resolution multi-sensor signal processing algorithm is to achieve high resolution performance without increasing N or the multi-sensor aperture.
Algorithms which belong to this category:
- have, asymptotically, infinite resolving capability - as the observation interval (i.e. number of snapshots) tends to infinity.
- suffer from the presence of system and/or modelling "uncertainties" - thus, eliminating the uncertainties (calibration) is an important problem for the operation of these algorithms.
Application areas include: Radar Signal Processing (e.g. SAR, MIMO), Acoustic Signal Processing, Single and Multi-Beam Array Signal Processing.
- [C1] Auto-Calibration
- [C2] Arrayed MIMO RADAR
- [C4] Low SWAP Target Localisation and Spatiotemporal Beamforming
- [I4] Multi-Beam SAR
- [O06] Source Separation for Electronic Surveillance
- [O07] Signal Processing Techniques to Reduce the Clutter Competition in Forward Looking Radar
- [O09] Low-Complexity Adaptive Beamforming Algorithms Based on Low-Rank Decompositions and Set-Membership Filtering
- [O10] SAR processing with zeros
- [O13] Joint Blind Enhancement and Passive Source Localisation of Acoustic Signals