Research On Time Synchronization Of Industrial Wireless Networks Based On Bayesian Estimation

Industrial wireless network is another research focus of industrial control field after the field bus technology, and it is a revolutionary technology to reduce the cost, expand application area of industrial measurement and control systems, so it is a new growth point of industrial automation products in the next few years. WIA-PA(Wireless Network for Industrial Automation-Process Automation)networks with independent intellectual property rights is an industrial wireless network technology that researched and developed by the Chinese Academy of Sciences, Southwest University and other 10 institutions in the support of National High Technology Research and Development Program of China.Time synchronization is an important component of every distributed systems, so it is an important support network service of WIA-PA networks that mainly has a TDMA (Time Division Multiple Access) access control method, playing the decisive role about the success of timeslot communication between nodes. Existing wireless sensor network time synchronization algorithms are all based on broadcasting time synchronization packets, because of the WIA-PA Mesh network portion has the restriction of that only two neighbor nodes can communicate in one timeslot, these existing algorithms can not well meet the WIA-PA network time synchronization requirements.Therefore, this paper in connection with the feature of WIA-PA network that inter-cluster timeslots are all dedicated timeslot, only can communicate with one neighbor node in one timeslot, gives the solution that set up logic synchronization tree of Mesh of WIA-PA networks based on the FTSP(Flooding Time Synchronization Protocol) algorithm. Then transfer the time synchronization packets along the logic synchronization tree accomplish the purpose that avoiding a large number of redundant packet transmission and reduce the cost of energy in time synchronization overhead. At the same time, to further improve the accuracy of time synchronization, Bayesian Estimation has been introduced to time synchronization algorithm in this article.Using Bayesian Estimation to reduce the error of estimation data of transmission latency time of synchronization packets in the network and the measurement error of local time,then do linear fit using optimized timestamp data, so the precision of time synchronization between nodes has obviously improved. This paper firstly introduced the main three types of industrial wireless network technology that being researching and developing. Then introduced the definition of the logic clock of nodes, analysis and compare several wireless sensor network time synchronization mechanism and the typical synchronization algorithms.On this basis, proposed the thinking of that meet the time synchronization requirements of WIA-PA networks through improving FTSP algorithm. Firstly, according to the characteristics of WIA-PA networks that Mesh router device only can communicate with one neighbor router node in one inter-cluster timeslot, build up the logic synchronization tree of Mesh network on the basis of that a Mesh node is needed to forward the adding requests when a new node want join the Mesh of WIA-PA networks, and synchronization packets are only sent to the neighbor router nodes which are the children nodes in the logical synchronization tree, so solute effectively the problem that the transmission of a lot of useless redundant synchronization packets, reduces the costs of transmission overhead in time synchronization.In addition, because of the feature of that the error of transmission delay estimation of synchronization packets and measurement error of local time of nodes meets the normal distribution, the paper improves the accuracy of transmission delay estimation data of synchronization packets and measurement data of local time of nodes through using Bayesian Estimation in the algorithm. After calculating the linear regression of relationship between the two nodes clock using optimized timestamp data, so the linear fit may reflect a more realistic relationship between two nodes clock. After learning of the relationship of clock between the nodes, one node can adjust its local time dynamically during the synchronization period to ensure that the time deviation between local time and the reference node time within a small range. This paper simulated the proposed FTSP-Bayes time synchronization algorithm using NS2 simulation software. Viewing from the simulation results, the algorithm solves the problem that the transmission of a large number of redundant synchronization packets in FTSP algorithm, reduces by 50% the number of synchronization packets at least, and also improves 2 u s on the precision of time synchronization between nodes. The time deviation between nodes may still stably remained at a relatively small range.Finally, the author takes part in the research and development program of WIA-PA networks. To solve the problem that the time synchronization among network nodes, the time synchronization algorithm FTSP-Bayes which proposed by this paper has applied in a WIA-PA network prototype system that built up. The test data of time deviation of between network nodes shows that the algorithm has a outstanding performance in the time precision when applied in WIA-PA network, the time deviation between the network nodes is less than 3.7μs. Through the monitoring of network operating conditions, It shows that the algorithm can satisfy the requirement of time synchronization of WIA-PA network,providing a solid foundation for the reliable operation of the network.