Improved source correlation estimation in wireless sensor networks

Emerging wireless sensor networks (WSNs) aim to overcome current needs for additional backbone infrastructure in mobile networks. Confronted with enormous requirements in terms of spectrum and energy efficiency, WSNs have been shown to benefit from distributed source coding (DSC) by means of multi-route redundancy. However, DSC in WSNs yields low performance regarding the correlation estimation of messages sent across different routes. Thus, we introduce a new dynamic threshold for the employed DSC technique, which minimizes the mean-square error (MSE) of such estimations. Moreover, we enhance the DSC algorithm using an over-determined linear equation-system such that the standard deviation is considerably smaller. The thereby proposed correlation estimation improves both signifying regions of the bit-error rate (BER) curve. The waterfall region is shifted by up to 10.5 dB and the error floor region is lowered by 70%. Additionally, we derive a BER boundary for the error floor which serves as a general performance indicator of the employed DSC technique.