An Uncertainty Quantification methodology for Broadband Wireless Access systems

Abstract

A Broadband Wireless Access (BWA) system comprises a Base Station and a group of M users in its vicinity, simultaneously serving N active users, N∈{0,1,2,…,M}, in Point-to-Multipoint (PMP) fashion. Depending on the concepts of system information and system entropy (i.e., mean information) we promote an Uncertainty Quantification (UQ) methodology for BWA systems. We evaluate the measures: system entropy (which represents the uncertainty, inherent to any system), average number of simultaneously active users, system resource utilization and availability, and average throughput per user, with respect to the parameters: system utilization and system size. Entropy is studied through its both components: nominal part and additional part. In information linear BWA system the additional part of entropy vanishes. Moreover, this methodology offers an entropy-based procedure for system utilization parameter estimation, showing that Maximum Likelihood Method (as an estimation procedure) and Fisher Information (as an uncertainty measure) can be successfully applied only to the information linear BWA system. Thus, the change of BWA system features is inherently built in this approach through the change of entropy additional part. Further, using the function of relative risk, given the active user average number, we quantitatively show that smallest uncertainty is associated to the Binomial BWA system (representing a private network), greater to the Erlang loss BWA system, and greatest at all to the information linear BWA system (both representing public networks), where the quantity N obeys Binomial, truncated Poisson and truncated geometrical distribution, respectively.