Cooperative evolution of SVM-based resource allocation for 5G cloud-radio access network system with D2D communication

Abstract

In fifth generation (5G) communication networks, the cloud-radio access network (C-RAN) is a new technique, where baseband processing units are decoupled from remote radio heads and a remote cloud-based centralised pool of baseband processing units is established. However, as the system's capacity grows, managing interference between cellular users (CUs) and device-to-device (D2D) users becomes a critical issue. This paper proposes a multi-class classification resources allocation scheme based on the cooperative evolution of support vector machine (SVM) to assign macrocellular users (MUs) spectrum resources to remote-head users (RUs) and D2D pairs, allowing sub-channels to be reused without compromising quality of service (QoS). First, the key resource allocation sets are determined, and a C-RAN resource allocation model is created. Because CUs and D2D nodes are allowed to access the same sub-channel, the ultimate challenge is described as a many-to-one matching sub-game. The 5G C-RAN system allocates resources via a cooperative evolution of an SVM-based multi-class classification algorithm based on user position estimates, with intercell and intracell interference utilised to build the training dataset. The training dataset is prepared based on intercell and intracell interference. Finally, the results show that the proposed cooperative evolution of SVM-based resource allocation approach outperforms standard resource allocation methods.