HSB-D2DStream: A Hybrid Semantic–Bit Communication Framework for Multi-User Live Video Streaming Over D2D Networks

With the explosive growth of mobile video services and live streaming applications, future wireless networks face increasing pressure to deliver high-quality multimedia content under limited bandwidth and dynamic network conditions. Device-to-device (D2D) communication offers a promising solution by enabling direct content exchange among nearby users and alleviating backhaul congestion. However, existing D2D caching and streaming schemes often neglect the stringent delay, jitter, and throughput requirements inherent to video transmission, as well as the unique characteristics of semantic communication (SemCom), where meaning fidelity depends on knowledge matching between the transmitter and receiver. To address these challenges, this paper proposes HSB-D2DStream, a Hybrid Semantic–Bit D2D Streaming Algorithm for multi-user live video delivery in cache-enabled D2D networks. The base station (BS) selectively caches video chunks to a subset of users, which then disseminate content to others through optimized multi-hop D2D links. Each user dynamically selects between semantic and bit-level transmission modes based on its required video quality and background knowledge alignment. The proposed algorithm jointly optimizes user clustering, cache node selection, and D2D topology to maximize overall Quality of Service (QoS) while satisfying constraints on delay, jitter, and error rate.