List of questions

From the complex motions of a marine vessel at sea, can the roll motion be isolated?

A roll sensor device has been built, but it works well in calm or smooth surface water only. It is necessary to use the device in mixed seas of varying sea states, and to eliminate the additional movements of heave and sway in order to isolate the roll motion.
Preliminary tests have already been carried out at the Edinburgh wave basin.

What level of edge processing is appropriate for resilient persistent surveillance networks?

We have a need to provide multimodal surveillance networks both for civilian and military security purposes. The overall purpose of the networks is to provide knowledge to military or civilian controllers to co-ordinate responses to events.
The sensor modes may be varied. For the purposes of the task, high bandwidth (such as video) and low bandwidth (such as acoustic) sensors should be considered.
The infrastructure may vary from pre-existing networks e.g. in a modern city, to non-existent e.g. in a rural area of operation where the entire infrastructure needs to be provided. The former may be wired; the latter may need to depend on batteries, generators and wireless connections
The key functions required of the surveillance network include:
• Coverage – ideally covering the whole area of interest without blind spots
• Persistence – the network should continue to operate 100% of the time for a pro-longed period
• Flexibility – the network should be scalable and or able to evolve coverage
• Resilience – damage to part of the network should be contained and have minimal impact on the rest of the network. Single points of failure e.g. a single control room should be avoided.
• Maintenance – other than if damaged the network should require minimal maintenance e.g. battery replacement
• Size, weight and power. These should be minimised within equipment
• Value for money – the network should be affordable
The concept is to include intelligence in the network so that data is only transferred when it is needed. The intelligence is devolved towards the edges so that events or behaviours are recognised near the sensors and only data or knowledge of value is transmitted. This is particularly important for bandwidth limited networks to avoid important information being swamped by routine data. Data may be stored near the sensor.
The two extremes of processing architecture are:
• centralised, where raw data is transmitted to a central processing hub;
• and extreme where all of the processing is conducted within the sensor unit.
Between these two extremes is a spectrum where the processing is de-centralised to local hubs or split between multiple granularities of the network.
As processing costs continue to fall, the cost equation will modify.
Therefore the question relates to the trade-off between the intelligence distribution options. The question should be addressed for several deployment times i.e. what is applicable for networks deployed in 2018, 2023, 2030?

Recognising whether a signal is a MIMO transmission or not

The problem of determining whether a signal is a MIMO transmission with little apriori knowledge has received little attention in the literature. The problem could be treated as a blind channel estimation problem, but this would generally require some other additional information, for example type of modulation used. Are there other ways of distinguishing a MIMO transmission ?

Methods and techniques for the detection and characterisation of phase coded maximal length sequences, below system thermal noise floor, of non-precisely known parameters.

Radar signals were classically high energy, short duration, events where range resolution and detection performance were determined by the classic pulse parameters. More modern systems change this model by utilising modulated waveforms to attain the required range resolution and increased duration (up to CW) to attain target detection from lower peak transmitted signals.

Sparse reconstruction of 3D scenes from multi-look Synthetic Aperture Radar

"SAR imaging in a 3D environment, for example airborne imaging of an urban scene, suffers from a problem referred to as layover, where objects at different heights (e.g. the tops of buildings and the ground) are imaged in the same pixels. If images are formed from multiple passes at different heights (but not enough passes to do a full 3D SAR), the pixels will have different combinations of contributions. It is reasonable to assume that the number of 2D surfaces contributing to layover in a region will be small. However, we expect significant, distributed reflections from most parts of the 2D surfaces. Can sparse reconstruction help to reconstruct the scene in 3D and, if so, to what degree might be possible?"

Distributed algorithms for TDOA for UAV swarms

The use of UAV swarms is now becoming a possibility, especially for surveillance applications. Low SWAP sensors will allow TDOA-based geolocation but this will requiring the use of suitable algorithms which may be best implemented using distributed processing techniques to share the computational load.