Protocol Design And Model Optimization Of Time Synchronization In Wireless Distributed Networks

Distributed wireless networks are multi-hop and self-organized systems composed by a series ofmobile nodes which have completely equal status. It mainly includes the wireless ad hoc networks,wireless sensor networks and wireless mesh networks. Distributed wireless network timesynchronization means that all the nodes in the network adjust their clock time using a distributedapproach, in order to achieve a unified time in the whole networks. The realization of timesynchronization can ensure that each node in the network collaboratively complete a variety ofdistributed tasks such as periodically sleep, TDMA based medium access control, data aggregationbetween nodes and so on. Therefore, how to efficiently and accurately realize time synchronizationfor the whole network has become a hot research question in the academic circles at home and abroad.This paper focus on two kinds of time synchronization protocol, that is mutual time synchronizationprotocol and master-slave time synchronization protocol. They are widely used in wireless ad hocnetworks and wireless sensor networks, respectively. The innovation points of our work aresummarized as follows.(1) Firstly, a revised beacon frame completion mechanism based on the existing mutual timesynchronization protocol is given. This mechanism can ensure at least one successful transmission ofthe beacon frame in each beacon period through a kind of fully distributed method. Then, we point outthat the value of backoff parameters W has an important effect on the performance of mutual timesynchronization. Finally, according to the process of mutual time synchronization, we propose atwo-dimensional Markov chain model to analyze the quantitative relationship between the time tosuccessfully complete the transmission of beacon frame and the value of W. Through the model, weobtain the value of backoff parameter W which makes optimal performance of the mutual timesynchronization in each beacon period. Comparisons between the numerical results derived by themodel and the simulation results obtained by EXata simulator validate the effectiveness of the model.(2) After the introduction and comparison of three kind of existing master-slave timesynchronization protocols, we firstly point out that the time synchronization protocol based on twicemessage exchange has wide application foreground. Afterwards, we analyze the shortcoming of TPSNin dealing with synchronization message conflict and propose an efficient collision avoidancemaster-slave time synchronization protocol. This protocol requires all the nodes form a tree structure,and the master node initiates the time synchronization process of each synchronization period through broadcasting req_syn message. According to the responses process of sub-node, we propose atwo-dimensional Markov chain model to analyze the relationship between the time to successfullycomplete the responses of synchronization message and the value of backoff window. Through themodel, we can get the optimal value of backoff parameter W with a given number of sub-nodes. Afterreceiving all the sub-nodes’ responses, the father node broadcasts rep_syn message, thus achieving thetime synchronization of sub-nodes. The number of synchronization message during eachsynchronization period is greatly reduced by the adoption of the optimal backoff parameter. Whendetecting the channel conflict, parent node inform it’s child nodes by broadcasting col_syn message,thereby improving the success rate of time synchronization. The effectiveness of the model and theprotocol is validated through comparing the numerical results of the model and the simulation resultsin EXata.