Accurate Energy-Aware Workload Distribution for Wireless Sensor Networks Using a Detailed Communication Energy Cost Model

Wireless Sensor Networks (WSNs) have gained a lot of attention from the research and industrial communities. In almost any application involving WSNs, energy efficiency is a primary concern. As the complexity of the motes and the applications increases (e.g., in-network DSP processing, internet-of-the-things), the energy of communication and computation processes have to be considered and optimized simultaneously. In this work, we propose an offline workload distribution approach based on integer linear programming (ILP) to reduce the energy consumption and extend the network lifetime for cluster-based WSNs. It takes communication and computation energy into account, employs a novel and detailed model for the communication cost, and provides optimal partitions for both symmetrical and asymmetrical networks. The experimental results confirm that the novel communication model can be applied to both CDMA and TDMA based MAC protocols with high accuracy. The estimated communication cost (approx. 10% deviation) is more accurate than the one employed in the previous works on partitioning (over 85% deviation). The incorporation of the accurate communication model into the workload distribution problem produces better partition result that reduces the energy cost by 16.8%. The economized execution time (less than 1 second) makes the ILP approach feasible for typical WSN applications and guarantees the optimality of the energy-aware partition.