Research On Reliability And Stability Of Application Layer Multicast

With the increasing deployment of broad techniques,media multicast has become one of the major network applications. However,due to the insufficient support from the network infrastructure, large-scale ip-multicast is difficult to be widely deployed. To solve this problem, researchers have proposed application layer multicast(ALM) for media multicast,in which unicast channels are built among ordinary end-systems to distribute the streaming data.In the dozens of application layer multicast models haved been purposed, high performance, scalability,reliability and stability have been key issues researchers have interest. On the one hand, end-system reliability level is far lower than network equipment like routers.End-system or link failures impact on the reliability of ALM.Tree-like multicast model exists the possibility of a single point of failure. In addition, on the other hand, from the large-scale level, with the heterogeneity and dynamic character, end-systems’ abrupt departures will becomes major factor in the instability. Reliability and stability of multicast services are particularly important for service providers.This paper focuses on reliability and stability of application layer multicast, and researches on overlay multicast architecture, clustering and constraints, the media and network coding, expansion of backbone and auxiliary. We design an application layer multicast simulator to verify effectiveness of the RSM models and mechanisms in reliability and stability. We have expanded the popular CDN-based IPTV technology by ALM and designed a ALM-CDN-hybrid HMALM-IPTV system.It can improve reliability and stability of system performance by structure expansion and coding mechanisms.We have analyzed the reliability and stability of current popular application layer multicast. The impact of reliability problems that include model structure limit, node failure, node performance, link failure, data errors and lost encoding mechanism. Stability problem of node failure is the main factor.Error rolerance and network optimization can improving the stability of models.We propose a series of formal definitions of application layer multicast clustering and constraints of structured.Based on it, we propose regular-balanced structured overlap network (RSM) and additional layer extension mechanism., We show that the multicast network can achieve much higher optimizing speed through dynamically clustering nodes with similar performance records. hierarchical structure makes it has better scalability and fault isolation capabilities.The structural characteristics of the node group to ensure balanced access to the degree exit data in the node joins a lower outage probability, the smaller the disturbance the system has excellent reliability and stability. Hybrid Mesh Application Layer Multicast (HMALM) can ensure reliability and stability by architecture.Then,we designed and implement a general event-driven application layer multicast simulator, through the application layer multicast traffic (Traffic), protocol, link, transmission, reliability and stability of the node and other key factors of flexible configuration, multicast protocol to achieve cross comparison test.A series of tests show that the hybrid model has a strong network multicast reliability and stability.Focusing on existing CDN-based IPTV system lack of scalability, we design a hybrid mesh-base model (HMALM-IPTV system) and the corresponding SVC spatial random linear coding scheme. Through these works we can achieve with the CDN-IPTV system resources sharing and scalability.Take, take full advantage of the cumulative hierarchical coding characteristics of SVC, to achieve the basic layer and enhancement layer encoding multiple concurrent transmission and buffer management. According to the network environment, End nodes select the appropriate receiving and decoding broadcast packets, which improved overall reliability and stability in general.Based on user behavior characteristics,we design a two-phase commit strategy. The end node make service level agreement with center node and obtain its license.Then, it execute the node join algorithm to join the system. Signing of the program to achieve business and operational management, can eliminate some of the side effects of free riders, with better security and scalability. According to heterogeneity of end system, priority scheduling strategy can enhance multicast system stability when users are joining or leaving the system.Application layer multicast reliability and stability of the Internet television (IPTV) and other large-scale application development has great theoretical and practical value. The hybrid network multicast model and two-phase commit are designed to affect the reliability and stability of user behavior to ensure the overall application layer multicast system reliability and stability.Reliability and stability of Application layer multicast is still an open research issue.In the future,we will concerned about the following topics:(1) What is the user behavior characteristic and how to affect the reliability and stability of multicast system?(2) How to surveil, control and maintain the reliability and stability of ALM system?(3)How to take full advantage of ALM to optimize current network resources?...