Simple and efficient network-aware user association rules for heterogeneous networks

Abstract User association (UA), the process that maps a user to a base station, is known to be an important problem in heterogeneous cellular networks. We study an online UA framework where association decisions are made for users individually to optimize an α-fair network utility function. We derive simple rules in the form of closed-form formulas that maximize the network utility for different choices of the fairness parameter α and are, in that sense, “network-aware”. These rules exhibit a “natural” notion of load that we identify. They require little information from the network while providing significant performance gains compared to conventional rules. We compare our rules to an upper-bound on the system performance which is obtained by recomputing UAs for all users in the system upon every change in the system state. We evaluate the performance of our rules using an in-house simulator and quantify their performance gap from the upper-bound. We, then, propose a way to reduce this gap by allowing periodic individual re-associations. We further evaluate the performance of our rules using NS-3 where none of the processes other than UA are under our control. The results confirm the superior performance of our rules compared to the conventional ones.