Signal processing and sensor control research is led by Professor William Moran. The team investigates problems associated with the control of sensing systems.
Associated problems including bandwidth allocation and distributed computing for networks of cooperating sensors. We work across topics including:
- information theory
- control theory
- target tracking and data fusion
- communications and
- signal processing
Our philosophy is to determine new fundamental results and then use these to guide the development of practical solutions to practical problems.
The team has supervised a number of PhD students and hosts visiting international research scholars and students.
Sensor Signal Processing
The recent technological advances in sensing hardware require sophisticated processing techniques to take full advantage of their capabilities. We make use of information theory, statistical signal processing and physics to develop algorithms to extract useful information for the detection, characterisation and recognition of objects in noisy environments.
Topics of particular interest in sensor signal processing are radar waveform scheduling, radar imaging and phase unwrapping.
Adaptive Sensor Networks
Advances in sensor technologies, computation devices and algorithms have created opportunities for significant improvements to building a coherent picture of an environment over large areas. Unfortunately, as information requirements grow, conventional network processing techniques require ever-increasing bandwidth between sensors and processors and exponentially complex methods for extracting information from the data. To address these problems it is necessary that sensing and computation be jointly engineering.
We take the perspective that a sensor system is a, possibly distributed, collection of sensing components with limitations imposed by power, bandwidth, computing resources, placement and other physical constraints.
This sensor system collects measurements from an evolving scene, and over time and within its limitations, adapts its performance to the scene to optimally extract information from the environment. The focus is on the development of adaptive sensor scheduling and sensor management algorithms.
Situation awareness is the process of identifying and quantifying threats to the attainment of one's objective.. Until the advent of automation, situation awareness was considering a product of training and experience. Recent advances in sensors and networking means operators are now confronted with large volumes of data that may change rapidly. Drawing rapid inferences from such large masses of data now exceeds human capabilities. Thus, there is a need for reliable, automated inference under uncertainty — this is the principal requirement of situation awareness.
Our work in this field is designed to deliver robust, automated inferential support tools to the human decision-maker, thereby enabling the focus to be on the most demanding issues.
Target tracking is a fundamental tool for any single or multi-sensor surveillance system. Such sensor systems, e.g. radars or sonars, report measurements from many diverse sources, only some of which are from the objects of interest. Target tracking algorithms must be capable of detecting, locating and often identifying these objects of interest. In the case of a multi-sensor system, tracking algorithms must also be capable of registering the data from each sensor system and fusing it to create a single, coherent picture.
Our work in this field is on designing target tracking algorithms that are both computationally efficient and accurate.
SLAM and control of UAV under network constraints
Advances in hardware have enabled small Unmanned Aerial Vehicles( UAVs) to be built with great flying controllability and reliability. In particular, various low cost, commercially available UAVs already possess the capacity to perform Simultaneous Localisation and Mapping (SLAM) tasks at small-to-medium scales. The development of software systems for autonomous navigation, sensor-based signal processing, integrated flight control, target tracking and object localisation, etc. is crucial to the success of the SLAM applications with UAVs.
Information Geometry of Sensor Networks
A better understanding of placement, planning and scheduling issues can be gained by exploring the connections between information geometry and performance of sensor networks for target tracking.
Firstly, the integrated Fisher Information Distance (IFID) between the states of two targets is analysed by solving the geodesic equations and is adopted as a measure of target resolvability by the sensor. The IFID and the well-known Kullback-Leibler divergence (KLD) are quite different though they measure ‘distance’. It is shown that the energy functional, which is the `’integrated, differential’ KLD, relates to the other distance measures.
Secondly, the structures of statistical manifolds may be demonstrated by computing the canonical Levi-Civita affine connection and Riemannian and scalar curvatures.
We show the relationship between the Ricci curvature tensor field and the amount of information that can be obtained by the network sensors.
For enquiries contact firstname.lastname@example.org