Probabilistic Offload Scheme in Integrated Cellular WiFi Systems

WiFi offloading is a cost-effective way of alleviating the problem of highly congested cellular networks. In this paper, we consider WiFi offloading problem in an integrated cellular WiFi system consisting of mobile base station (MBS) and WiFi access point (AP). We propose a probabilistic offloading scheme, where cellular packets that arrive at the queue of MBS are offloaded to the queue of WiFi AP with an offload probability. The offload probability is determined to minimize the average delay experienced by the cellular packets while guaranteeing stability of both cellular and WiFi system. We model the arrivals and fulfillments of data services of a cellular operator as an M/M/1 queue. We investigate two priority disciplines: First-In-First-Out (FIFO) and Non-Preemptive Priority Rule (NPPR). We provide an exact optimal offload probability for FIFO, but present an upper bound on the optimal probability for NPPR. Numerical investigation is used to verify the optimality of the proposed solutions, to examine the effect of packet arrival rate of MBS and compare the average delays for FIFO and NPPR.