Distributed source coding in large wireless sensor networks

We study the performance of distributed source coding in large wireless sensor networks obtained with enhanced correlation estimators. Distributed source coding is especially useful when data correlation exists since it tries to remove the redundancy in the information; and dense sensor networks are rich in correlations. Existing results from information theory show that this compression can be executed in a distributed fashion and without any performance loss in comparison with the centralized approach. However, there is still performance gap between the theoretical bounds and the results achieved with practical implementations. In order to mitigate this, we propose the use of enhanced correlation estimators. Simulation results show a performance improvement in the energy consumption by reducing the number of transmitted bits compared to classical methods.