Research On Overlay Network QoS Technology For Distributed Multimedia Applications

Distributed multimedia application has three fundamental characteristics: network resource sharing, real-time implementation, and transparently intercourse, these makes it requests more complicated and more sensitive QoS(Quality of Service). However, the current Internet network can only provide Best-effort Service, it cant be used for distributed multimedia application to providing reliable QoS assurance due to three facts that:1) Lack of scalability of resource scheduling mechanism and can not guarantee fair distribution of sharing resource;2) Lack of disaster protection and recovery mechanism and can not guarantee stable system implementation;3) Lack of communication guarantee mechanisms and can not adapt to heterogeneous network transmission. Therefore, bearing real-time network multimedia practice is still a grim challenge for Internet. IP QoS can provide service assurance mechanism based on IP network layer, such as Integrated Service, Distributed Service, MPLS and so on, but because of expensive extending cost and compatibility problems of routers, these technologies can not be used in mass application. Overlay network has been attached by most researchers for its advantage of improving distributed network application service of quality without changing fundamental network structure. It provides reliable service routing protocol through constructing virtul service network at the application layer. Therefore, we can apply overlay network into distributed multimedia system for achieving important break. To solve the problems resulted by the allocation of resources, disaster recovery and transmission protection, this dissertation researches on QoS of overlay network. The paper’s work is supported by Scientific and Technological Project of Wuhan City (Urban Public Security Emergency Networking Platform) and Education Research Plan of Guangdong Province (No.2006D90104005). Firstly, an Overlay Service Network Model (OSN) is discussed, then based on the OSN model, three theories of technology research including a fair load balancing QoS routing algorithm, a fast QoS re-routing algorithm, a QoS-aware dynamic method, are proposed. Finally, To solve the problems of the bottleneck of the service quality in surveillance applications, an overlay-based video surveillance system architecture is designed. And we develop a series of city networked video surveillance project based on the theories. In this paper, the research results from the perspective of the application layer of highly efficient and reliable exploration QoS control strategy provide scientific guidance of pioneering ideas for large-scale distributed multimedia systems designed. It is of great theoretical significance. And it is also important for research and development of video surveillance systems, video conferencing systems, IPTV and other useful products.In this paper, the theoretical algorithm, application programs and innovative contribution to have the following(1) A fair load balancing QoS routing algorithm of overlay networkTraditional load balancing routing algorithm to choose minium cost of resource consumption of the path of access to services in order to achieve a balanced distribution of load resources, but did not consider the resources of the local path to the bottleneck problem caused by network congestion. To solve this problem, this paper puts forward a fair load balancing QoS routing algorithm. This method is the introduction of the degree of characterization of the path of the resource bottleneck fairness factor and the cost model to establish a new load-balancing utility function, and uses adaptive weighting system load status of amendments to the weight of QoS constraints. Simulation results show that the algorithm can reduce the probability of emergence of a bottleneck link, as compared with the traditional algorithm on average 2%-5% of the QoS service success rate, reducing by 1%-5% of the resources of the fluctuations in the load rate of saturation.(2) A fast QoS re-routing algorithm of overlay networkTraditional QoS re-routing algorithms used the back-up method of reconstruction and restoration of interrupted service to switch the path, but did not consider the routing delay of the reconstruction costs of the process, which will cause delay in recovery time and low efficiency of service problems. To solve this problem, this paper puts forward a fast QoS re-routing algorithm. The method is the introduction of reconstruction delay of the path to restrict at alternative path selection. It is in favor of election of the resumption of faster alternative path. Expand of a Failure regional approach will be to elect the course of a single alternative path into the path of a collection of selected options. It will improve success probability of re-routing. Simulation results show that the algorithm can effectively improve the routing efficiency of recovery time, compared with the traditional algorithm to reduce the average 5%-8% of the restoration of the delay time, increase 5% increase success rate of repair.(3) A dynamic QoS-aware method of overlay networkTraditional QoS-aware overlay network determine the changing needs of the abnormal based on static QoS critical threshold. However, the methods need to rely on the experience of inferred threshold range, affecting the measurement accuracy. And it can not be applied to a variety of different circumstances of the situation-aware QoS parameters. To solve this problem, this paper puts forward a dynamic QoS-aware method based on Gause Mixture Model. The method establishes a unified system of evaluation model through self-learning training in the principles of the samples, Mask the differences between the parameters of the ability to adapt to different QoS parameters for monitoring. Simulation results show that this method can effectively detect the continuity of QoS performance anomalies and to avoid unexpected error detection of abnormal samples to improve detection accuracy. Actual network environment tests show that the algorithm can detect more than 92% of the unusual circumstances and to control the false detection rate of 6%.(4) An overlay-based video surveillance system architecture modelTraditional video surveillance system will demand a high quality of service of multimedia services and low quality of service the needs of the general level, lack of scalability of the scheduling mechanism for QoS services. It can not satisfy the mass of heterogeneous quality of service in surveillance application. To solve this problem, this paper puts forward an overlay-based video surveillance system architecture model. In the model, single-layer system architecture is extended to the two-tier service-oriented architecture. A virtual multimedia service resources overlay network with the deployment of QoS routing protocol services. The method can solve QoS-guarantee problem in surveillance application. The practical application cases prove the effectiveness of the system model.In summary, to solve the problems of distributed multimedia application, this dissertation researches on QoS of overlay network. Four technologies are proposed, including a fair load balancing QoS routing algorithm, fast QoS re-routing algorithm, QoS-aware dynamic method and overlay-based video surveillance system architecture model. The researches solve the problems:the bottleneck resources, the efficiency of fault-tolerant recovery time, QoS-aware inaccurate, the mass of heterogeneous quality of service in surveillance application. Lastly, the dissertation summarizes the research achievement and looks into the future research.