Resource allocation for uplink OFDMA C-RANs with limited computation and fronthaul capacity

This paper considers the joint fronthaul resource and rate allocation for the OFDMA uplink cloud radio access networks (C-RANs). This amounts to determine users' transmission rates and quantization bit allocation for I/Q baseband signals, which must be transferred from remote radio heads (RRHs) to the cloud over the capacity-limited fronthaul network. Our design aims at maximizing the system sum rate through optimal allocation of fronthaul capacity and cloud computation resources. Toward this end, we propose a novel two-stage approach to solve the underlying non-linear integer problem. In the first stage, we relax the integer variables to attain a relaxed problem, which is solved by employing a pricing-based method. Interestingly, we show that the pricing-based problem is convex with respect to each optimization variable, which can be, therefore, solved efficiently. In addition, we develop a novel mechanism to iteratively update the pricing parameter which is proved to converge. In the second stage, we propose two different rounding strategies, which are applied to the obtained continuous solution of the relaxed problem to achieve a feasible solution for the original problem. Finally, we present numerical results to demonstrate the significant sum-rate gains of our proposed design with respect to a standard greedy algorithm.

I. Introduction

C-RAN has been recently considered as an architecture evolution for the next-generation wireless network. By realizing various communications and processing functions in cloud, C-RAN enables more efficient utilization of computational and radio resources, which results in better network throughput as well as reduced network deployment and operation costs. To realize these benefits, one has to address many technical challenges, e.g., to efficiently utilize computation resource in the cloud, fronthaul capacity, and to design suitable communication schemes in the access network [1].