The Research Of Pre-processing Of Data Synchronization Protocol SyncML

Mobile devices have been inextricably linked with people's lives.A lot of data are generated on mobile devices every day,and how to keep data consistency between different mobile devices becomes an important research subject.Among such more protocols of data sync,SyncML Protocol is an industry standard protocol for implementing mobile data sync,more and more data sync applications based on SyncML are in use,but these apps will take a long time and even fail to sync when synchronizing large amounts of data.Therefore,in this case,how to reduce sync time has become the focus of the current mobile data synchronization.At present,many scholars have improved the SyncML protocol to improve the synchronization efficiency,which are mostly focused on the optimization of the network transport process.Such as the Huffman coding method for file synchronization was supposed to reduce the amount of sync transmission,through the compression of sync data.And also some research are based on transaction-priority synchronization model,through the optimization of the transport process to reduce the time.Although these studies can improve the efficiency of synchronization,it has neglected the consideration of the classification of files' synchronization before they transfer.This paper argues that,in the case of a huge number of sync files,the improvement of the transport process of the protocol is important,but the pre-processing optimization of files before transport is equally worthy of careful study.SyncML do not distinguish a large number of files before synchronization,and do complex operations with all files,such as recording change log,setting up sync anchor and so on,which leads synchronization to be slow and even failed.To solve this problem,by careful analysis of scenarios,this paper has proposed the concept of Modification Frequency Degree and has divided the files into two categories: Non-Frequently Modified Files and Frequently Modified Files.For the latter,synchronization will obey the original methods.For the former,the complex operations such as recording common change log will be ignored in order to save expense,especially in the scenario of a large number of files.In addition,according to the proposed improved method,we have modified SyncML sync protocol and represetation protocol,and then we have discussed more details for implementation and its universal applicability.To verify the modified SyncML protocol can indeed effectively reduce sync time,the paper implemented an iOS sync client based on SyncML protocol.By synchronizing the data files in the Documents document within iOS application sandbox,it also designed an experiment to compare the sync time for which original protocol's Refresh Sync from Client only and the Backup Sync proposed in the paper.The experiment used four sets of different number of files as the control group on three different mobile devices,then comparing and analyzing the sync time obtained within the experiment.The experiment results show that the sync efficiency between the Backup Sync and the Refresh Sync from Client only is almost same when synchronizing little files.But with the increase of the files and the proportion of those not required to modify frequently files,the Backup Sync spent less time than the Refresh Sync from Client only every time.The conclusion is that in the case of a large number of files,especially in the case that the proportion of the Non-Frequently Modified Files is large,the Backup Sync proposed in the paper spends less time than the sync type proposed in the original protocol.