Dynamic resource management in competitive wireless data networks: a game theoretic framework

Summary form only given. The competition among wireless data service providers brings in an option for the unsatisfied customers to switch their providers, which is called churning. The implementation of wireless local number portability (WLNP) is expected to further increase the churn rate (the probability of users switching the provider). However, the existing resource management algorithms for wireless networks fail to fully capture the far-reaching impact of this unforeseen competitiveness. From this perspective, we first formulate non-cooperative games between the service providers and the users. A user's decision to leave or join a provider is based on a finite set of strategies. A service provider can also construct its game strategy sets so as to maximize their utility (revenue) considering the churn rate. Based on the game theoretic framework, we propose an integrated admission and rate control (ARC) framework for CDMA based wireless data networks. The admission control is at the session (macro) level while the rate control at the link layer packet (micro) level. Two admission control modes are considered: one-by-one mode and batch processing mode in which multiple users are admitted at a time. We show that: (i) for the one-by-one mode, the Nash equilibrium in pure strategy can be established for both under-loaded and fully-loaded systems; and (ii) for batch processing mode, there is either an equilibrium in pure strategy, or a dominant strategy exists for the service provider. Therefore, the providers have clearly defined admission criteria as outcome of the game. Users are categorized into multiple classes and offered differentiated services based on the price they pay and the service degradation they can tolerate. We show that the proposed ARC framework significantly increases the provider's revenue and also successfully offers differentiated QoS to the users.