Welcome to Coventry University website

Team: Dr James Brusey, Dr Elena Gaura.

Although fault detection and management are currently receiving increasing attention within the WSN community, much work focuses on solutions for specific types of faults and errors or specific WSN applications. When more generic solutions to the WSN robustness problems are sought, it is cumbersome, if not impossible to combine such existing, specific techniques into a WSN system component or service that could handle multiple types of fault and is applicable to any-size/any-type wireless sensor network.

Past fault-related research at Cogent and discussions with both WSN users and application domain specialists have confirmed that a fault management framework able to handle a variety of faults (ranging from sensing device faults through networking and communications to systems software faults) is needed to enhance the applicability of the WSN technology and increase end-users' confidence.

Hence, the aim of this project is to develop a fault management framework for WSNs in order to improve robustness, reliability and wider adoption of WSN technology. The research addresses a need for a holistic approach that tackles faults at a number of different protocol stack layers with low overhead in terms of computing bandwidth and energy consumption.

Towards this aim, a number of research questions are currently being answered:

• Is it possible to define a fault management framework that is generic and suited to various WSN applications requirements?
• What information could be gained from multiple protocol stack layers to accelerate the fault detection process?
• How do the application requirements affect the decision-making during the fault recovery process?
• How should the framework be designed to fulfil the requirements of WSNs, such as node energy-efficiency, balanced communication workload and process workload sharing among the nodes?

The major constraint throughout the framework design process is the overall energy stored in the wireless sensor network, whilst the design objective is to produce a network with extended life and reliability concurrently maximizing the quality of the information collected and delivered by the network.

Currently the issues of network degradation management and resources sharing are being addressed in order to extend the network's operational life in the context of an engine health monitoring application. Additionally, several framework components have been developed including a method of collaborative sensor self-diagnosis, which shows considerable advantages in efficiency and false alarm rate when compared with other diagnosis methods.