A New Adaptive Compression Scheme for Data Aggregation in Wireless Sensor Networks

Wireless sensor nodes typically have limited processing capabilities and are powered by batteries. The amount of energy expended in transmitting a single data bit would be several orders of magnitude higher when compared to the energy needed for a 32 bit computation. Thus, to maximize network lifetime, data transmissions should be minimized without losing vital information. In this paper, a novel adaptive compression scheme using nonlinear estimation theory is proposed for data aggregation. Satisfactory performance of the proposed compression scheme in the presence of noise, distortion, and quantization errors is demonstrated using Lyapunov approach. The proposed scheme is contrasted with existing compression schemes using various metrics applicable to wireless sensor networks such as energy efficiency, distortion and compression ratio. Simulation and hardware experimental results demonstrate almost 50% energy savings with very low distortion (less than 5%) and overhead. By iteratively applying the proposed scheme at the cluster head nodes, higher energy savings are obtained with a tolerable level of distortion.