UBC Theses and Dissertations
Energy-efficient resource allocation and cooperation in wireless heterogeneous networks Ramamonjison, Rindranirina
The deluge of mobile data demands a drastic increase of wireless network capacity. A heterogeneous network design, in which small cells are densely deployed, is required to satisfy this demand. However, it is critical that this dense deployment does not lead to a surge in energy cost. The aim of this thesis is to design energy-efficient resource allocation methods and explore the value of cooperation in terms of energy cost. In particular, three different cooperation schemes are studied. First, a multi-cell coordination scheme is proposed for maximizing the energy efficiency of heterogeneous networks. Although this problem is not convex, convergent algorithms are devised to find an efficient power allocation. We found that this simple coordination can offer a significant energy efficiency gain even in dense networks. Second, a joint energy allocation and energy cooperation is proposed for heterogeneous networks with hybrid power sources and energy storage systems. For this study, an offline optimization problem is considered, in which the cells allocate their energy over time based on average rate contraints, the changing channel conditions and the fluctuating energy arrivals. It is found that an optimal use of the harvested energy significantly improves the energy efficiency. A much larger gain is obtained when energy cooperation is also leveraged, i.e. when the cells can exchange their harvested energy through a smart-grid infrastructure. Third, the trade-off between energy cost and performance is addressed for cooperative clustered small-cell networks. In this cooperative model, the small-cell base stations form a cluster of distributed antennas to collectively serve their mobile users. Hence, a joint optimization of cell clustering and cooperative beamforming is proposed to minimize the total energy cost while satisfying the users’ quality of service. The problem is formulated as a mixed-integer convex program and solved with a decomposition method. For a given clustering, a distributed beamforming algorithm is also designed to achieve near-optimal performance at a small cost of signaling overhead. Through simulations, it is shown that these algorithms converge fast and enable the cooperative small cells to save valuable energy.
Item Citations and Data
Attribution-NonCommercial-NoDerivs 2.5 Canada