Long-term fairness in multi-cell networks using rate predictions

Providing fair service to users in cellular networks is of significant importance to mobile operators. However, the geographic variability of network coverage and irregularity of user demand in space and time has made consistent fair service provisioning a challenging problem. Therefore, vehicular users traversing small cells experience significant service fluctuations. In this paper we investigate how rate predictions of mobile users can be utilized to improve long-term fairness. We use the α-fair utility function to formulate a predictive long-term resource allocator that improves fair user service over multiple cells. The α-fair utility provides the flexibility to obtain the desired long-term tradeoff between fairness and throughput. Numerical results indicate promising fairness-throughput gains over schemes that do not incorporate user rate predictions.