Research On RPL Routing Protocol Based On Multi-sink

Wireless Sensor Network(WSN) is widely used, including military(intelligence interception, surveillance, targeting and detection), environment(monitoring and warning of fire, flood and pollution), health(health monitoring, medical management), home(Smart environment) and other commercial(interactive museum, inventory management and monitoring, vehicle tracking and monitoring). In 2008, Internet Engineering Task Force(IETF) quickly analyzes the routing requirements of typical applications in WSN with a detail. As the result, the IPv6 routing protocol for low power RPL network is newly proposed.WSN usually deploys a small number of sinks and lots of sensor nodes. Sensor nodes communicate with the sink node. Currently, researches on RPL are focused on testing performance and improving the network under a single sink node. However, the positive role of multi-sink nodes has been proven in various research papers.This thesis firstly analyzes the upstream and downstream routing mechanism, and outlines the theoretical basis of RPL. The tree cluster formed in RPL is called DODAG, the root of which plays the role of the sink node used to transmit network control packets(DIO). After receiving control packets, nodes calculate the RANK value, thereby forming the entire DODAG. Then, based on the open source software for RPL-Contiki RPL, this thesis analyzes the mechanism for multi-sink nodes in Contiki RPL and finds the following problems: First, changing frequently in DODAG will increase DIO packets transmission and affect the performance of the network traffic, especially the downstream traffic. Second, the loop mechanism has problem. Processing the INFINITE_RANK value of Rank in DIO message will cause the loop. Third, the node is not sensitive to the changes of neighbor state, which can lead to reselecting the best parent is delayed.Considering these three questions, this thesis explores solution to each question respectively. For frequent changing among DODAGs issue, this thesis sets a threshold method, thus reducing the transforming number between DODAGs and enhancing the successful reception rate of downward packet in network. For loops issue, this thesis proposes a loop avoidance mechanism, additional processing the INFINITE_RANK value of Rank in DIO message, thus reducing the number of loops and enhancing the successful reception rate of packets in network. For the issue the node is sensitive to the changes of neighbor state, this thesis presents neighbor initiative detection mechanism. The node maintains neighbor relations by sending NS packets to the best alternative at the right time, thus enhancing the performance of the network.