Automatic Parameter Optimization of Sensor Network MAC Protocols

Wireless sensor networks are concerned with resource and performance constraints. In the light of these constraints, sensor network protocols must operate efficiently and effectively at all times. This applies in particular to the medium access control (MAC) protocol, which dictates to a large extent the energy consumption of a sensor node by controlling the use of the radio transceiver. Moreover, it determines the timing of communication among neighboring nodes and thus per-hop latency and per-hop reliability. To achieve the best possible performance, the MAC protocol must be configured with appropriate parameters and automatically adapted to changes in network conditions and application requirements. This thesis presents a novel system for automatic optimization and adaptation of sensor network MAC protocols. It adapts the protocol parameters at runtime to ensure optimal performance and compliance with the application requirements. The optimization approach is based on constraint programming and optimizes multiple objectives simultaneously. The approach is applied in a case study to X-MAC [9], a flexible and widely used MAC protocol for sensor networks. Experimental results obtained from a small-scale network of real sensor nodes show that the proposed system keeps MAC protocol performance close to the optimum under varying network conditions and different application requirements.