Handover Performance of Priority Schemes in Cellular Networks

In this paper, we study the performance of three handover priority schemes - the pure guard channel method (GCM), GCM with first-in-first-out (GCM-FIFO), and dynamic priority queuing (DPQ) - and, for the first time, compare them for different scenarios and provide guidelines as to which scheme to use under which scenario. In the first scenario, the channel holding time is assumed to follow an exponential distribution. For this scenario, we derive new accurate formulas to evaluate the performance of GCM and GCM-FIFO schemes. Our results show that existing analytical frameworks for GCM and GCM-FIFO overestimate the handover failure probability. Recent field measurements have shown that the best fits to measured data for channel holding time are lognormal and mixed lognormal distributions rather than an exponential distribution. We quantify and compare the handover performance of GCM, GCM-FIFO, and DPQ for ideal (i.e., exponential channel holding time distribution) and for practical (i.e., lognormal and mixed lognormal channel holding time distributions) scenarios. Our results show that, depending on the cell type, the number of channels per cell, and the traffic load, the operator should use either DPQ or GCM. For example, for high handover traffic and high traffic load, the operator should use GCM-FIFO; for light-to-moderate traffic load and a high number of channels per cell (40 or more), DPQ seems to be the best option. To make the handover scheme adaptive, a switching mechanism can be used such that, depending on the time of the day, traffic load, number of channels per cell, and cell type, the handover scheme that provides the best performance is selected. The guidelines provided in our paper are, in most cases, valid for both exponential and pure/mixed lognormal distributions.