In wireless communications, merging data from various distributed receivers is a common strategy adopted to improve overall link performance. In this paper, we address the problem of optimally merging bit decisions from various distributed receivers when no information about the channels or about the performance of the receivers is available. Herein, we derive a novel algorithm to blindly form the maximum likelihood estimates (MLEs) of the individual bit error rate (BER) of each of the distributed receivers. We show that the variance in the estimates decays at the same rate (upto a constant) that would be achieved with perfect knowledge of the transmitted bits. Subsequently, we use these estimates to optimally fuse the individual bit decisions, thereby improving overall performance. We show that the above fusion algorithm asymptotically achieves the performance of the globally optimal fusion rule. Also, simulation results show that in all practical operating regimes, this empirical fusion rule outperforms the standard "majority fusion rule" receiver and also the best (minimum BER) receiver in the bank.
展开▼
