Efficient resource management for vod streaming service in wireless networks

Abstract

Providing VoD streaming services over wireless networks is very challenging due to high QoS requirements of VoD streaming services and the limited capacity and error-proneness of the wireless environment. So, this dissertation proposes efficient resource management schemes to for VoD streaming services over wireless networks.First, we propose a flexible forward error correction (FEC) and a fair selection scheme of scalable units that utilize a layered coding structure of H.264/SVC. Three error-resilient techniques (e.g., unequal error protection, FEC, and retransmission) are adapted to minimize the total distortion of VoD streaming service. For flexible FEC, a rateless FEC code is adopted. The FEC code rates are based on the possible number of retransmission, the condition of the wireless channel and the layered coding structure of H.264/SVC for each packet. A theoretical study is performed to show how to utilize the possible number of retransmission for an adaptive FEC code rate. With fair selection, regular and retransmission-requested packets are not differentiated. They compete for resources without fixing the retry limit. Thus, excessive retransmission is prevented and the proposed scheme effectively provides capacity-limited and delay-constrained VoD streaming services. For this fair selection of scalable units, we formulate the problem using binary integer programming and propose an effective low complexity selection algorithm based on a priority index. The priority index of a scalable unit considers time-varying resource requirements for transmission, as well as the impact on video distortion. The proposed algorithm prioritizes packets according to the priority index and the H.264/SVC structure. We show that the proposed scheme can minimize the total video distortion compared to other heuristic procedures. Other effects of the various factors are also considered for the performance of the new scheme.Second, we propose two batching schemes to overcome t...