Scalable Convex and Non-Convex Optimization for Dense Wireless Networks

Abstract

The evolution towards the next generation mobile networks is characterized by an unprecedented growth of smart devices. This will inevitably result in drastic data avalanches beyond the capability of current systems. One promising solution is Cloud radio access network (Cloud-RAN), this thesis starts with investigating a CSI acquisition approach in dense Cloud- RANs. We propose to convert the available spatial and temporal prior information into appropriate convex regularizes. Further, we adopt the alternating direction method of multipliers (ADMM) algorithm to solve the resultant large-scale high-dimensional channel estimation problem. After that, this thesis exploits the topology control in dense Cloud-RANs. Specifically, we propose a sparse and low-rank optimization approach for network topology control in the partially connected Fog-RAN. This model helps find the network topologies with the maximum number of allowed connected interference links. To address the coupled challenges, we propose a smoothed Riemannian optimization framework by exploiting the quotient manifold geometry of fixed-rank matrices, followed by a smoothed sparsity-inducing surrogate. The proposed algorithm has a much lower computational cost compared with state-of-the-art matrix factorization parameterized methods. Simulation results further demonstrate the appealing sparsity and low-rankness trade-off in the proposed model, thereby guiding the network deployment in dense Fog-RAN. In the second part of thesis, we investigate our earlier work in WSNs and propose a novel joint design of sensor nodes clustering and data recovery. Furthermore, we take both the energy-efficiency and data forecasting accuracy into consideration and investigate the trade-off between them. Simulation results demonstrate that our joint design outperforms the existing algorithms in terms of energy consumption and forecasting accuracy.