Fast Distributed Multiple Access Based Selection With Imperfect Parameter Knowledge

Several wireless systems that seek to exploit multiuser or spatial diversity need to opportunistically select a ‘best’ node from among the many available nodes. This is done on the basis of a suitability metric that is locally known at each node, and is initially not known elsewhere. While a fast, scalable, multiple access based algorithm is known today for doing this in a distributed manner, it assumes perfect knowledge of statistical parameters such as the cumulative distribution function of the suitability metric and also the total number of nodes. In this paper, we develop a novel and comprehensive analysis of the algorithm when the parameter knowledge is imperfect – as is always the case in reality due to statistical inaccuracies or the time-varying nature of the system. We show that imperfect knowledge does negatively influence the performance of the selection algorithm, and must be accurately accounted for.