Joint Network-Channel-Correlation Decoding in Wireless Body Area Networks

The performance of a novel non-binary joint network-correlation channel decoder (NB-JNCCD) is examined for wireless sensor networks with correlated sensor data. This work is motivated by the example of a wireless body area network that gathers measurements of various physiological variables, such as blood pressure and body temperature, that are frequently correlated. In order to exploit the correlated sensor data, a novel correlation decoder is developed under the assumption that decoder knows the first order pairwise correlation between any pair of sources. Simulation results show that the proposed correlation decoder performs similarly to a system in which the correlation between sources is removed prior to transmission using source coding techniques. However, unlike source coding, the correlation decoder does not require costly sensor-to-sensor communication and thus is more suitable for wireless sensor networks.;The correlation decoder is then integrated into a previously proposed non-binary joint network-channel decoder in order to create the NB-JNCCD. Simulation results demonstrate that the NB-JNCCD outperforms previous decoders that combine channel and network decoding, but that do not exploit correlated sensor data. The performance of the NB-JNCCD is further studied via computer simulation in order to understand the effects of various systems parameters on its performance. This includes different network topologies, different channel models, and variations in the channel code rate and pairwise source correlation. Overall, the proposed NB-JNCCD along with the accompanying analysis provides a method for designing efficient wireless sensor networks for transmitting correlated data.