Flexible Resource Allocation in 5G Ultra Dense Network with Self-Backhaul

Abstract

Self-backhaul is flexible and cost efficient for ultra dense network (UDN) since it provides backhaul using the same wireless technology as access link. Content prediction and caching can lower the load on backhaul and improve user experience. Therefore, an algorithm named tri-stage fairness (TSF) is proposed to solve the resource allocation problem in UDN with self-backhaul and caching, in which cells without direct network connection (rTP) access core network through donor TP (dTP). In TSF, rTP determines to transmit files cached in rTP (rTP files) or the files not cached in rTP (dTP files) according to delay and link capacity, and allocate access link resource using proportional fairness algorithm. dTP allocates backhaul resource among its users and rTPs with fairness considerations, and decides the time each rTP spends on backhaul link. Fairness, efficiency, overhead and complexity are jointly considered in TSF. To facilitate system level simulation, a traffic model considering the influence of caching is also introduced. Simulation results suggest flexible resource allocation between access and backhaul link yield substantial performance gain.