Research On File Fragmentation And Data Prefetching Strategy In Video-on-Demand System Based On P2P-CDN

The rapid development of Internet and the increasing usages of smartphone have led to a rapid growth of video traffic.The video service mainly includes video live broadcast and video on demand(VoD)service.The VoD service is one of the most important streaming media applications on Internet.When the users of VoD system increase,the burden of VoD servers increases rapidly in the traditional VoD system with client/server(C/S)architecture,the capability of this system decreases,and the bottleneck severely limits the rapid development of the system.Content Delivery Network(CDN)technology and Peer-to-Peer(P2P)technology have been applied to the VoD system,which might solve this problems.However,the VoD system with CDN has the disadvantages of poor scalability and high maintenance cost,and the P2 P system has the difficulty of node management.To solve these problems,this thesis proposes a VoD system based on P2P-CDN using Spring Cloud microservices framework,which reduces the complexity of system and strengthens the communication ability.With the integration of P2 P and CDN,this system effectively reduces the pressure on the core server,and improves the scalability.By analyzing the advantages and disadvantages of the existing file fragmentation algorithms,an equal-size fragmentation algorithm for convertible transmission units is proposed.This algorithm not only has the advantage of easy deployment,but also can adaptively change the transmission unit according to the network condition,which can ensure that the fragmentations to be reliably transmitted in different network environments.The experimental results show that the fragmentation algorithm can ensure high fragmentation efficiency and adapt to different network transmission environments compared with the existing typical fragmentation algorithms.To improve the performance of this system further,this thesis optimizes the data-prefetching strategy based on location satisfaction.The multiple-attribute decision-making(MADM)theory has been used to model the selection of intervals of prefetching data blocks(i.e.,anchor interval)and the technique for ordering preference by similarity to an ideal solution(TOPSIS)algorithm to solve the MADM.Through the optimization,the optimal data prefetching strategy is obtained.We name this strategy based on the playback delay and positioning satisfaction as the DSA strategy.The simulation results show that the DSA strategy can guarantee a lower playback delay,and can further increase the positioning satisfaction with less extra bandwidth.This shows that the DSA strategy has a good optimization effect on data prefetching.