Multiple-Input Turbo Code for Joint Data Aggregation, Source and Channel Coding in Wireless Sensor Networks

In wireless sensor networks, the data collected by neighboring sensor nodes show high correlation. To exploit the data correlation for reducing the amount of data while transmitting data from sensor nodes to their base station, this paper introduces multiple-input turbo (MIT) code to implement a joint source coding and data aggregation. If there exists an explicit communication between two sensor nodes, their correlated data bits are first interleaved with each other in accordance with their correlation relation, and then are encoded and punctured. On the other hand, if there is no direct communication between sensor nodes, the data sequences are encoded with respect to side information based on the distributed source coding principles. Whenever the bit error rate needs to be improved, MIT code is used for channel coding as well. MIT code employs partial interleaving to reduce energy consumption and memory size requirements for even small-size information blocks. The simulation results show that the bit error rate performance of MIT code outperforms turbo codes to some extent, even if MIT code implements partial interleaving on interleavers.