A Multi-Level Strategy for Energy Efficient Data Aggregation in Wireless Sensor Networks

In this paper, we have proposed energy efficient multi-level aggregation strategy which considers data sensing as continuous stochastic process. Our proposed strategy performs filtration of sensed data by removing the redundancy in the sensed data pattern of the sensor node using Brownian motion. Further, the filtered data at the sensor node undergoes entropy-based processing prior to the transmission to cluster head. The head node performs wavelet-based truncation of the received entropy in order to select higher information bearing packets before transmitting them to the sink. Overall, our innovative approach reduces the redundant packets transmissions yet maintaining the fidelity in the aggregated data. We have also optimized the number of samples that should be buffered in an aggregation period. In addition, the power consumption analysis for individual sensors and cluster heads is performed that considers the communicational and computational cost as well. Simulation of our proposed method reveals quality performance than existing data aggregation method based on wavelet entropy and entropy based data aggregation protocols respectively. The evaluation criteria includes-cluster head survival, aggregation cycles completed during simulation, energy consumption and network lifetime. The proposed scheme reflects high potential on practical implementation by improving the life prospects of the sensor network commendably.