Evaluating mobile phone handoff behavior using chi-square statistical test

Mobile networks reuse frequency bands based on a color map to increase the capacity of the network. A handoff should occur when a mobile unit moves from the influence of one base station with weaker signal into another's that has stronger signal. Handoff behavior of all units is an important factor in quality of service of a mobile phone service. Handoff decisions, also called mobility decisions, are made by mobile phone based on the observed power from base stations. Premature, delayed or exceedingly sensitive decisions are considered poor decisions. Excessive poor decisions result in degradation of service quality in otherwise a healthy mobile system. Conventional research focuses on improving hand-off algorithms. Most of the published work on verification of effectiveness of handoff algorithms is analytical or focuses on data collected under pristine laboratory conditions. A unit that makes good mobility decisions, theoretically or in the laboratory, may not behave as expected in the real world however. We propose a process of evaluating hand off behavior using large amount of diagnostic phone data collected in the real world that is used for identification of adverse trends or aberrant behavior of various models. In this paper, we discuss a chi-square statistical test to evaluate the performance of specific mobile unit model by comparing the behavior of a test mobile unit against a well- established behavior profile. If the behavior of the test model deviates significantly from the well-established profile, it is considered deficient in its handoff behavior that deserves further analysis. The test was developed in such a way that a large amount of units can quickly be tested. The same test can be used to compare performance of all mobile phones in one region to performance of same mobile phones in other regions. Furthermore, our test is useful in determining difference of handoff behavior when the mobile units are moving in opposite directions.