Research And Implementation Of Reliability Assurance Method Of Routing-Oiented Publish/Subscribe System

WSN (Web Service Notification) is a set of publish/subscribe system standard developed by the OASIS organization, defined a Web services specifications through the use of topic-based publish/subscribe model for notification. Subscribers send a subscription request to the message producer. When message producer receiving the request, he will create a subscription resources and return the service address of the resource. When a message producer produces a notification message matching a subscription, he will send the notification to consumers who have subscribed this topic. The WSN proposed agent-based message forwarding, in this scenario, the subscribers send subscription requests to the agent, and the producers send notifications to agent. Then it is the agent to determine forwarding the notice to which consumers. In this way, producers and consumers do not need to know each other’s existence, which plays the role of the decoupling.WSN-based publish/subscribe system designed an independent administrator in addition to the system topology, as the administrator of does not make publish/subscribe operations, but as a topology manager to help establish the topology, maintenance the topology, query and configure the node information. The administrator is the premise of the system start-up, and all other nodes start-up should interact with it to get the configuration information and group information. Notification messages of the system are routed to subscribers, and all subscribers interested in the same topic form a subscription tree. The subscription tree determines the routing efficiency and reliability. With more and more different message types and different delay requirements, higher requirements are proposed to assurance system reliability.This paper designs a method to ensure the system reliability WSN-based publish/subscribe system. It guarantees reliability improvement from several aspects, such as master-slave dual-machine backup, the priority queue based on Bloom Filter, routing algorithm improvement based on B+tree and so on. Dual-machine backup administrators know each other’s existence through heart messages. They synchronize system key information timely, and backup administrator replaces the primary administrator’s function after its failure. Priority decision engine based on Bloom Filter can determine the priority of messages according to their characteristics. Messages with priority can be forwarded out through multi-level priority queue. This could deal with messages of different delay sensitivity. The new routing algorithm based on B+tree instead the original binary tree algorithm and guarantee efficient forwarding with large-scale topological structure. Through these aspects, system reliability can be powerfully improved.