User Association and Scheduling Based on Auction in Multi-Cell MU-MIMO Systems

We study the user association and scheduling problem in multi-cell multi-user multiple-input multiple-output systems with dynamic traffic. Two successive linear beamforming schemes, namely, zero forcing-successive interference cancellation and adaptive orthogonal beamforming, are adopted in the uplink channel and the downlink channel, respectively. The confidence level of signal-to-interference-plus-noise ratio of a random user is analyzed, in order for base stations (BSs) to decide on the maximum number of active users while guaranteeing the worst-case quality of service. We also derive the conditional expected rate of a random user knowing its local channel state information, in order for users to evaluate the performance and select the BS. Noticing that the performance can be divided into different classes, we model the user association and scheduling problem as an auction game. The auction is equipped with a strategy-proof pricing scheme and can be conducted in a distributed fashion with a limited information exchange. Shown by the simulation, the proposed mechanism has the dominant performance in respect of social welfare, throughput, and load balancing.