QoE‐aware Q‐learning resource allocation for NOMA wireless multimedia communications. Issue 5 (22nd September 2020)

• Record Type: Journal Article
• Authors: He, Shuan; Wang, Wei
• Is Part Of: IET networks. Volume 9:Issue 5(2020)
• Page Start: 262
• Page End: 269
• Publication Date: 2020-09-22
• Subjects: multi‐agent systems -- quality of experience -- multimedia communication -- multi‐access systems -- resource allocation -- learning (artificial intelligence) -- mobile radio -- synchronisation -- telecommunication computing -- telecommunication congestion control; Q‐QoE resource allocation framework -- QoE‐aware Q‐learning resource allocation -- multimedia flow -- NOMA user equipment -- communication resource cost -- reinforcement Q‐learning -- wireless channel states -- multiple agent reinforcement learning problems -- sequential single agent reinforcement learning approach -- nonorthogonal multiple access -- SARL scheme -- hybrid control traffic -- quality of experience -- NOMA wireless multimedia communications -- QoE gain; Computer network architectures -- Periodicals; Computer network protocols -- Periodicals; Information networks -- Periodicals; Telecommunication systems -- Periodicals; 004.605
• DOI: 10.1049/iet-net.2020.0021
• Languages: English
• ISSNs: 2047-4954

Abstract

Abstract : Hybrid control traffic and multimedia flow over emerging non‐orthogonal multiple access (NOMA) could provide both very low latency and very high bandwidth. In this study, a Q‐learning‐based resource allocation scheme is proposed to improve the quality of experience (QoE) for NOMA user equipment (UE) in downlink wireless multimedia communications. In the proposed framework, the utility is modelled as the QoE with regard to communication resource cost, where UE acts as the agent in the reinforcement Q‐learning. UE observes the wireless channel states and takes resource allocation actions based on the immediate reward of QoE gain and communication cost. In addition, benefiting from the NOMA communications, the authors propose to solve the multiple agent reinforcement learning problems with the simplified sequential single agent reinforcement learning (SARL) approach. The numerical simulation results demonstrate the efficiency of the proposed Q‐QoE resource allocation framework and prove that the UE would obtain desirable QoE performance with the SARL scheme.