Stochastic packet collision modeling in coexisting wireless networks for link quality evaluation

Abstract

The packet delivery ratio (PDR) performance of a wireless communication network interfered by a coexisting network is determined by the collision-time. In this paper, we propose a stochastic collision-time model for coexisting wireless networks, and analyze it in particular for coexisting low-rate wireless personal area network (LR-WPAN) and WLAN. Although, the packet collision-time of the interfered network is a complex process, as it depends on the packet size and packet inter-arrival time distributions of both networks, its properties can be inferred by modeling the interference as an alternating renewal process, leading to a stochastic collision-time model. The proposed collision-time model is utilized to derive the theoretical collision-time distributions for periodic, exponential and gamma inter-arrivals. The comparison of the theoretical and simulation based collision-time distributions for the studied inter-arrivals suggests that the proposed collision-time model can be used for performance analysis of coexisting wireless networks. In order to investigate the effects on the collision-time distribution in a realistic multi-terminal WLAN traffic shaped by the CSMA/CA mechanisms, heavy tailed hyper-Erlang and gamma distributions are fitted to an experimental inter-arrival times data-set. The goodness-of-fit tests of both distributions show that the gamma distributed inter-arrivals model the observed interfering traffic good enough, which allows analytical evaluation of LR-WPAN link quality using the derived collision-time model.