A pragmatic analysis of user mobility patterns in macrocellular wireless networks

Abstract

Due to multifaceted human behavior, synthetic models are inept at realistically modeling long term mobility characteristics of users. The diversity in mobility character adds yet another dimension to this complex problem. Empirical studies are essential and are capable of providing realistic user models. This paper examines the real-time mobility traces of users and identifies key mobility parameters, which are used to classify users and create homogenized groups. Based on mobility and degree of predictability, a mobile user classification is attempted. As per-user mobility management schemes proposed in the literature are difficult to implement, it is essential to adopt a class or group based approach to facilitate implementation of dynamic schemes. Further, this paper characterizes in-session and out-of-session Cell Residence Time (CRT), the feature that critically influences several management tasks. The out-of-session CRT distribution has been represented using a heavy tail distribution. The applicability of the model for various classes of users has been studied. The results of this study can be used to spawn a more realistic user model, for simulation studies.