Improved Low-Complexity Fusion Receivers for Synchronous DS-CDMA Sensor Networks

We present two new low-complexity Bayesian-fusion linear receiver structures for synchronous non-orthogonal DS-CDMA wireless sensor networks. The first receiver treats the underlying hypothesis as a system parameter and estimates it to arrive at a global decision. The second receiver is similar to the MMSE multiuser detector, except that it accounts for the covariance between the sensors´ transmitted bits in the system model. Both receivers correctly exploit the covariance among the local decisions to achieve improved performance. We evaluate the receivers´ performance in the presence of local-sensor and channel errors due to multiple-access interference (MAI), Rayleigh fading, and noise and show that their performance exceeds that of all of the suboptimal receivers proposed in S.K. Jayaweera (Jan 2005).