User-demand-aware wireless network selection: A localized cooperation approach

We study the network selection problem where multiple users with diverse user demands compete for access in wireless networks. Most existing network selection algorithms commonly suffer from the low efficiency of the social welfare, particulaly for distributed optimization approaches. Centralized optimization approaches can improve the efficiency, but they may incur much cost in network architecture, signaling, and computational complexity. We harvest the diverse user demands across users and propose a local improvement algorithm (LIA). Different from centralized approaches or distributed approaches, the key idea behind the LIA is introducing localized cooperation into networks who share users, called coupled network pair (CNP). Exploiting the spatial distribution of networks, the proposed algorithm decomposes global social welfare optimization into subproblems with low complexity, where each CNP cooperatively reassociates users with user demand awareness. Under a novel game formulation, we proved that the LIA can achieve promising performance. To speed up the convergence of the algorithm, we further exploit the spacial independence among CNPs and propose an enhanced LIA. Finally, simulations indicate that the proposed algorithms achieve much better performance with relatively short convergence time, compared with three distributed algorithms.