Equitable Extended Reality via Cooperative NOMA in Decentralized Mobile Edge Computing Systems

• Published in: Conference record - Asilomar Conference on Signals, Systems, & Computers pp. 694 - 700
• Main Authors: Reifert, Robert-Jeron, Dahrouj, Hayssam, Sezgin, Aydin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.10.2024
• Subjects: Computers; Delays; Extended reality; Multi-access edge computing; NOMA; Performance evaluation; Relays; Resource management; Scalability; Vectors
• ISSN: 2576-2303
• DOI: 10.1109/IEEECONF60004.2024.10942766

Extended reality (XR) applications pose strict technical demands on future communication systems in terms of rapid connectivity and resource-intensive task computations. To this end, this paper addresses the benefits of an optimized mobile edge computing system using battery-limited edge computers (ECs) so as to meet the intricate requirements of future XR applications. The paper classifies well-connected and poorly-connected XR devices as strong and weak devices, respectively. Such platform is then provisioned through the use of a cooperative non-orthogonal multiple access (Co-NOMA) optimized scheme. With the aid of device cooperation, the paper illustrates how Co-NOMA improves the weak devices' performance by flexibly selecting either the direct or relay transmission link. More specifically, to achieve equitable XR services, the paper maximizes the minimum rate among all XR devices by jointly managing computation, communication, and link selection resources subject to power consumption, computational capacity, and delay constraints. The proposed solution then utilizes discrete relaxations, successive-convex approximation, and fractional programming techniques, and is shown to be amenable for distributed implementation across the multiple ECs. Simulations illustrate the proposed approach efficacy, especially in terms of minimum rate and delay performance in decentralized XR systems, as compared to several state-of-the-art methods.