Video Streaming over D2D Networks

Abstract

Device-to-Device (D2D) communication has been presented as an innovation that can improve the cellular network performance by exploiting the proximity-based service between closely-located devices. Enabling D2D communication increases the energy efficiency, improves the capacity of the network and reduces the communication delay. Despite the above-mentioned advantages, D2D communication presents some challenges, for example, the need for proper interference management, power control, mode selection and device discovery. Nowadays, the increasing demand for video streaming has led to rapid growth in data traffic that is unable to be handled by traditional networks. Consequently, many works in the literature suggested employing D2D communication for video transmission to offload the cellular network and enhance the quality of video streaming. Moreover, the emergence of video-based applications has stimulated the need for high-performance D2D communication. This thesis focuses on video streaming over D2D communications underlaying a Long Term Evolution (LTE) network where Scalable Video Coding (SVC) is assumed. In particular, joint resource allocation, mode selection and power control for multiple D2D pairs are addressed. The objective is to maximize the throughput of D2D pairs while considering the minimum data rate requirements by both the Cellular Users (CUs) as well as the D2D pairs and maintain video quality and continuity. Resources are allocated to each CU and D2D pair in three modes of operation; cellular, dedicated and reuse and a mode selection algorithm is implemented. Furthermore, a packet-layer video assessment model is applied to predict the impact of network conditions on video quality. Finally, the effect of mobility on mode selection is examined. The performance of the proposed scheme is evaluated through extensive simulations and compared to the scenarios where only one mode of transmission is used for all D2D pairs. Simulation results show that mode selection improves the throughput of D2D pairs while providing better video quality. We assess the effect of user mobility on system performance and observe quality degradation for high mobility scenario.