Research On Caching Scheme In Time Shifted TV Systems

With the rapid development of computing technology and broadband networking and convergence of telecommunications networks, broadcast networks and internet, streaming media applications obtained more attentions. As a typical streaming media business, Time-shifted TV service is introduced to the market by internet service provider and broadcast TV service provider. Thus, the research on time-shifted TV also has academic and practical significance. Time-shifted TV provide broadcast TV service to users, and enables users to watch TV programs with a time shift and VCR-like controls like pause, jump, stop, fast forward, fast rewind etc operation. Due to high volume of data and real-time storage, time-shifted TV requires high network bandwidth and disk I/O bandwidth. The capacity of a single server is quite limited, so clustered servers are needed to provide large scale streaming media services.The work in this dissertation is based on "China High Performance Broadband Information Network" (3T'Net) project and "Hainan bidirectional HFC experimental network" projects. This dissertation discusses cache scheduling strategy for time-shifted TV server clusters, including distributed storage cluster, shared storage cluster and large scale cluster. For each kind of clustered server system, this dissertation gives an analytical model to describe the problems and finally proposed a corresponding policy. The main research contents and contributions are summarized as follows:1. The designing of distributed storage time-shifted TV cluster faces a cache placement problem of deciding how to cache channels to multiple servers so that the blocking probability is minimized subject to memory capacity constraints. We propose a popularity-based random placement (PRP) scheme together with the genetic algorithm (GA) to find an optimal or approximate optimal solution of the problem. The experiment results reveal that our proposed algorithm is efficient on improving the performance of time-shifted TV cluster in terms of minimizing blocking probability.2. To alleviate disk I/O bandwidth bottleneck of a shared storage cluster, a classification replication algorithm and a novel cache placement algorithm, named as Cost-aware Least Load Proportion First (CALLF) were proposed. It was formulated as a multi-objective optimization problem based on the time-based popularity model. The algorithms focus on the time-varying nature of Time-shifted TV, and explore previous stored information to reduce the cost of redeploying. We also present a dual-threshold reconfigure scheme to further balance the load of servers dynamically. Simulation reveals that the proposed algorithms are efficient on improving the performance of Time-shifted TV system in terms of total outgoing bandwidth requirement, and allow the tradeoffs between the bandwidth utilization and the reconfiguration overhead.3. In large scale time-shifted TV server clusters, by adopting cooperative caching technique, the free cache memory of all the servers in the cluster can be used more efficiently. A cooperative cache buffer scheduling strategy is presented, which makes use of each server to perform the caching task by manager server. It will help to reduce disk accesses, resulting in the improvement of the overall throughput of cluster. The proposed method is also designed to handle VCR-like operations. Simulations are performed to investigate its performance, and the results show that it can reduce disk I/O overhead and provide high throughput in clustered servers.