Research On Distributed Synchronizationtechnology In Dynamic Ad-hoc Network

With the advancement of wireless network technology,dynamic ad-hoc networks have been applied to more diversed fields.The increasing complexity of working scenarios usually require that the dynamic ad-hoc networks working with more complex collaboration,and in general time synchronization of the nework is the basis for ad-hoc networks to work collaboratively.The main feature of distributed synchronization compared to master-slave synchronization is that there is no master node that provides time services in the network,and all nodes have the same status.The node exchanges time information with its neighbors and adjusts the local clock.The network gradually realizes that the clock of each node converges to the average clock of the network to complete time synchronization.Since distributed synchronization does not have a fixed time information forwarding path,distributed synchronization can cope with topological changes and link instability in dynamic ad-hoc networks.Although distributed synchronization has strong robustness and can meet the time synchronization requirements in dynamic ad-hoc networks,too much time information interaction between nodes will still bring about the problem of excessive communication overhead.In this thesis,we will focus on the time synchronization in the dynamic ad-hoc network.Through the research of the existing distributed synchronization technology,a distributed synchronization protocol suitable for the specific dynamic ad-hoc network is proposed.At the same time,an optimization solution is proposed to reduce the excessive distributed synchronization overhead.The main points of this thesis include:(1)We research the existing time synchronization technology including master-slave synchronization and distributed synchronization,and compare the similarities and differences between the two synchronization methods.The focus is to study the scope of of the distributed synchronization and its specific synchronization process.We summarize the methods and existing problems of distributed synchronization by investigating the existing distributed synchronization technology.(2)Research the characteristics of dynamic ad-hoc network which is different from other wireless networks,and propose a distributed synchronization method suitable for dynamic ad-hoc network.In a dynamic ad-hoc network,there are not only problems such as topology changes and link instability,but also problems such as uneven distribution and dynamic changes of satellite timing functions in more complex scenarios.After investigation and research,a weighted distributed time synchronization method is proposed,which can deal with the uneven distribution and dynamic changes of satellite timing functions in a dynamic ad-hoc network.(3)Combined with specific scenarios,a new distributed time synchronization protocol in a dynamic ad-hoc network is proposed.Directional antennas can bring longer communication distances and higher data transmission rates for nodes in dynamic ad-hoc networks,but they also impose certain constraints on the design of time synchronization protocols.By studying the related technologies of directional antennas,using its characteristics,a distributed time synchronization protocol suitable for dynamic ad-hoc networks using directional antennas is designed.And use the Java language to simulate and verify the protocol.(4)A distributed time synchronization method based on reinforcement learning is proposed.The distributed synchronization method may bring a lot of communication overhead due to excessive interaction with neighbors.Through experiments,it is found that the synchronization of the network can still be established by selecting some neighbors instead of all to exchange synchronization information in the distributed synchronization.Combining the relevant theory of reinforcement learning,a distributed synchronization method is proposed,and the the method is verified through simulations.This method can effectively solve the self-adaptive trade-off between time synchronization effects and communication overhead.