Global Recovery Strategy Based On Heterogeneous Nodes And Adaptive Node Failure Detection Algorithm Research In Mobile ALM Environment

At present, the computing power and storage capacity of mobile intelligent terminal equipment has increased gradually while its price has died down progressively, which made mobile intelligent terminal equipment abundantly available. As a result the streaming media terminal application is becoming more and more popular. Providing smooth high quality streaming media service for mobile terminals has become a new research hotspot.The idea of application layer multicast relaying on terminal hosts responsible for data packet replication and forward and reducing streaming media server load stress has sprung up.The stability problem of application layer multicast tree nodes leaving, which forced children to adjust position in multicast tree and caused data stream suspended, seriously has affected the user's QoS. Therefore, it is necessary to enhance the stability of the application layer multicast tree, which is the inevitable requirement to improve customer satisfaction.Aiming at the application layer multicast nodes dynamic causing stability problem,this article has considered shortening nodes recovery time as the breakthrough point, and studied the recovery strategy of application layer multicast tree nodes and failure detection algorithm. The innovation points of this article are main performed as following two aspects:First of all,aiming at nodes initiative leaving application layer multicast tree,the paper has put forward a global optimized restoration algorithm based on heterogeneous nodes. The strategy is based on the cognition that members of the multicast nodes fast or slowly would leave the application layer multicast tree, which made it necessary to take enough measures for unpredictable nodes leave. Therefore, the application layer multicast tree has maintained a global node information table and provided controlled number candidate parent nodes according to the global information table. Aiming at nodes initiatively exiting, candidate parent nodes collecting key information of rejoined node such as time delay and degree and so on and optimized furtherly. Simulation results show that the algorithm has lower rejoin delay, at the same time is conducive to the stability of the layer multicast tree.Secondly, in view of nodes failure, the paper has put forward an adaptive and double timeout checking failure detection algorithm base on message delay prediction. Based on historical message delay, it can compute the next arrival time, at the same time the strategy of double timeout detection was taken, which can effectively judge. The experimental results and analysis show that the algorithm can achieve high detection speed and accuracy.