User Association Optimisation in HetNets: Algorithms and Performance

Abstract

The fifth generation (5G) mobile networks expect significantly higher transmission rate and energy efficiency than existing networks. Heterogeneous networks (HetNets), where various low power base stations (BSs) are underlaid in a macro-cellular network, are likely to become the dominate theme during the wireless evolution towards 5G. However the complex HetNets scenario poses substantial challenges to the user association design. This thesis focuses on the user association optimisation for different HetNets scenarios. First, user association policy is designed for conventional grid-powered HetNets via game theory. An optimal user association algorithm is proposed to improve the downlink (DL) system performance. In order to address the uplink-downlink (UL-DL) asymmetry issue in HetNets, a joint UL and DL user association algorithm is further developed to enhance both UL and DL energy efficiencies. In addition, an opportunistic user association algorithm in multi-service HetNets is proposed for quality of service (QoS) provision of delay constraint traffic while providing fair resource allocation for best effort traffic. Second, driven by increasing environmental concerns, user association policy is designed for green HetNets with renewable energy powered BSs. In such a scenario, the proposed adaptive user association algorithm is able to adapt the user association decision to the amount of renewable energy harvested by BSs. Third, HetNets with hybrid energy sources are investigated, as BSs powered by both power grid and renewable energy sources have the superiority in supporting uninterrupted service as well as achieving green communications. In this context, an optimal user association algorithm is developed to achieve the tradeoffs between average traffic delay and on-grid energy consumption. Additionally, a two-dimensional optimisation on user association and green energy allocation is proposed to minimise both total and peak on-grid energy consumptions, as well as enhance the QoS provision. Thorough theoretical analysis is conducted in the development of all proposed algorithms, and performance of proposed algorithms is evaluated via comprehensive simulations.