Autonomous Management and Control of Sensor Network-Based Applications

A central challenge facing sensor network research and development is the difficulty in providing effective autonomous management capability. This is due to a large number of control parameters, unexpected changes of the network topology, and dynamic application requirements. Network management is also a challenging task for the remote user due to the large-scale of the network and scarce visibility of live network happenings. Preferably the network should have autonomous decision-making capabilities as network conditions and application requirements change. In this paper, we consider Octopus, a tool that provides live information about the network topology and sensor data. The tool provides monitoring and visualisation functionality to the user. In the original version of the system, network state was controlled manually. This paper describes how Octopus has been reengineered to accommodate a multi-agent system to provide autonomic managing capabilities. In particular, we detail two distinct architectures, the static and mobile agent architectures, which can be effectively applied to deliver autonomous system management.