Distribution of Heterogeneous Traffic Resources for 5G Cognitive Radio Networks based on Cooperative Learning and QOE

Abstract

In addition to offering faster speeds and lower latency, 5G networks provide enhanced availability, significantly higher capacity, improved stability, and superior connectivity. The Mean Opinion Score (MOS), which quantifies the subjective quality of the user experience, has emerged as a widely accepted metric for assessing the quality of various types of network traffic. As we move towards the 5G era, measuring end-user quality has become increasingly important in the evolution of wireless communications. This paper presents a resource allocation approach for 5G cognitive radio networks that leverages cooperative learning and prioritizes Quality of Experience (QoE) for integrated heterogeneous traffic. The solution is based on a distributed underlay Dynamic Spectrum Access (DSA) system that utilizes MOS as a foundational metric for managing resource allocation across real-time video and data traffic, each with distinct characteristics. The proposed technique is designed to meet strict interference limitations for primary users while simultaneously maximizing the overall MOS. This is accomplished by employing reinforcement learning in a system where primary users and secondary users share the same frequency band of interest, ensuring efficient coexistence. Importantly, MOS serves as a standardized measure that allows for training across nodes carrying various types of traffic without compromising performance.