A game-theoretic analysis of wireless access point selection by mobile users

Abstract

A user located in a congested area of a wireless LAN may benefit by moving to a less-crowded area and using a less-loaded access point. This idea has gained attention from researchers in recent literature [A. Balachandran, P. Bahl, G. Voelker, Hot-spot congestion relief in public-area wireless networks, in: IEEE WMCSA, Callicoon, NY, June 2002; M. Satyanarayanan, Pervasive computing: visions and challenges, IEEE Personal Communications 8 (4) (2001) 10-17]. However, its effectiveness and stability are questionable. Each user selects the access point that offers the optimal trade-off between load and distance to be traveled. Since users are selfish, a user's selection may adversely impact other users, in turn motivating them to change their selections. Also, future user arrivals and exits may invalidate current selections. This paper presents the first game-theoretic analysis of this idea. We model access point selection as a game, characterize the Nash equilibria of the system and examine distributed myopic selections that naturally mimic selfish users. We analytically and empirically assess the impact of user diversity and dynamic exit patterns on system behavior. The paper contributes to a deeper understanding of the costs, benefits and stability of such a solution in various usage scenarios, which is an essential pre-requisite for real-world deployment.

Localized Access Point Association in Wireless LANs with Bounded Approximation Ratio

“…We adopted the number of users connected to an AP as the load capacity of this AP by assuming all users have the same minimum quality requirements. Mittal, Belding, and Suri [18] also adopted the same metric to model the load capacity of APs. Although Kelsen [16] proposed a local algorithm for parallel matching in expander graphs and proved that the proposed algorithm can find the maximum matching with high probability in sub-linear running time, it is unpractical in IEEE 802.11 based wireless LANs because it requires complicated communications between neighbor nodes and a large number of rounds to converge.…”

A Framework for Dynamic Network Architecture and Topology Optimization

“…In [25], a noncooperative game is used to model the association process to Wi-Fi-based APs. The cost function that each user aims at minimizing depends on the AP load and the distance the associating device needs to travel to be able to associate to the desired AP.…”

Designing a Practical Access Point Association Protocol

“…In this work we model the decentralized AP selection with selfish users as an extension of the weighted singleton congestion game in which the weight of a user varies as the associated AP changes. Other modelings of the wireless infrastructure selection include [18] and [19], targeting at different scenarios and goals. The major distinction of our work is that we design and implement on the commercial chipset an online greedy AP selection protocol in the distributed manner.…”