User Association in HetNets: Impact of Traffic Differentiation and Backhaul Limitations

Abstract

Operators, struggling to continuously add capacity and upgrade their architecture to keep up with data traffic increase, are turning their attention to denser deployments that improve spectral efficiency. Denser deployments make the problem of user association challenging, and much work has been devoted to finding algorithms that strike a tradeoff between user quality of service, and network-wide performance (load-balancing). Nevertheless, the majority of these algorithms typically consider simple setups with a single type of traffic, usually elastic non-guaranteed bit rate (GBR). They also focus on the radio access part, ignoring the backhaul topology and potential capacity limitations. Backhaul constraints are emerging as a key performance bottleneck in future networks, partly due to the continuous improvement of the radio interface, and partly due to the need for inexpensive backhaul links to reduce capital and operational expenditures. To this end, we propose an analytical framework for user association that jointly considers radio access and backhaul network performance. Specifically, we derive an algorithm that takes into account spectral efficiency, base station load, backhaul link capacities and topology, and two traffic classes (GBR and non-GBR) in both the uplink and downlink directions. We prove analytically an optimal user association rule that ends up maximizing either an arithmetic or a weighted harmonic mean of the achieved performance along different dimensions (e.g., uplink and downlink performances or GBR and non-GBR performances). We then use extensive simulations to study the impact of: 1) traffic differentiation; and 2) backhaul capacity limitations and topology on key performance metrics.