Distributed Multimedia Synchronization And Parallel Coding Techniques

In distributed multimedia system, the adoption of media server based on parallel architectures is the major method to deal with the increase of users and improve the QoS. Compared with single processor media server, the media server system is characterized by its complicated technological process and difficulty to maintain the temporal constraint relationships between different media objects; inappropriate scheduling algorithms will reduce the efficiency of the system; meanwhile when media server takes the adaptive source coding techniques, unsuitable parallel coding strategy will reduce the system performance. Therefore, this dissertation makes a study of the synchronization and constraint relationships in distributed multimedia system, the scheduling algorithm and the parallel video coding techniques of media servers. The content is mainly as follows:1,In order to describe fully the temporal constraint relationships of distinct medias in distributed multimedia system, we propose a new powerful synchronization model– MHPN (Multimedia Hierarchical Petri Net).Compared to other models, the priorities of MHPN include simplicity, clarity and powerful interactive controlling ability. These properties enable the dynamic characteristics of multimedia message in multimedia system to be described completely and also the temporal constraint and synchronization relationships among different multimedia objects to be described clearly and fully. In accordance with the distinct demands at different levels, MHPN is able to build hierarchical models.In light of the temporal constraint relationships and interactions demands between server and client of the distributed interactive multimedia system, we proposed a synchronization model: D-MHPN ( Distributed MHPN) which is based on MHPN and can uniformly describe the synchronization relationships and interactions in distributed multimedia system and provide more powerful abilities of message capitation and controls as well.2,Based on the performance analysis of scheduling algorithms of media server, a quality-oriented adaptive scheduling algorithm, which focuses on parallel media signal processing and transmission, is presented in this dissertation.When the media server providing adaptive QoS, the scheduling algorithms applied to sequential processing task are not suitable for parallel media processing, because that multi-nodes may be devoted to one same task. In accordance with the characteristics of parallel video processing and the requirements on mixed scheduling of sequential and parallel data processing tasks, the quality-oriented adaptive scheduling algorithm assign the parallel task to fewer computing nodes as far as possible. This leads to overhead reducing of data...