Selection based resource allocation for decentralized multi-user communications

Decentralized multiple access networks require dynamic spectrum allocation to efficiently and fairly allocate resources among multiple users. We view the allocation problem from the standpoint of diversity combining, and in particular as an explicit case of selection combining (SC). We consider a decentralized multi-user parallel-access spectrum allocation algorithm which generalizes SC. Each user selects the frequency band with the highest gain among the subset of bands not already being used by other users. We present an exact bit error rate (BER) analysis of the algorithm. We consider Nakagami-mm fading, and treat the cases of independent identically distributed (i.i.d.) channels, independent but not necessarily identically distributed (i.n.d.) channels, and correlated channels. In deriving the BER, we first present new results on the exact average error probability corresponding to the rrth order statistic, in each channel scenario. Numerical results show that the performance of the algorithm is close to that of the highly complex optimal search method, especially in the correlated fading case.