Energy and throughput optimization of a Zigbee-compatible MAC protocol for wireless sensor networks

This paper describes quantitative power and performance results from the simulation of an asynchronously scheduled MAC protocol for WSNs, called PE-MAC. It improves over past work by optimizing the transmitted power control, the back-off delay and minimum back-off exponential, depending on the sensed channel conditions. Retransmission is faster when the channel is good, and slower when it is either noisy or overloaded. The algorithm has been first modeled using Mathworks StateFlow, and functionally debugged at a high level of abstraction using the Mathworks simulation environment. Then we used automated code generation techniques to download an executable onto a power-aware simulator of a ZigBee-compatible node platform, and we analyzed separately the impact of each improvement we made over past work. We also compare the simulation results with the prediction of an abstract model of the PE-MAC algorithm, created using Markov chains.