Research On Data Aggregation Algorithm In VANET

With the development of times and the progress of the society,the vehicles play a more and more important role in our daily life,vehicular networks has become one of the most important medium for communicating among vehicles.However,the limitation of wireless channel bandwidth become the bottleneck of vehicular networks development with the increase of applications in.In this case,the data aggregation algorithm,which can reduce the data amount and relieve the bandwidth load effectively,gains more and more attentions by the researchers.The thesis propose the multi-application and low-overhead adaptive in-network data aggregation(MLAIA)on the basis of the existing data aggregation in the vehicular networks.Firstly,the algorithm presents a multiple applications oriented information broadcasting mechanism with low overhead.Each node is at the state of their own to maintain the operation of the whole system in the networks.The broadcast information from different information source will be aggregated to reduce the occupied network resources which has been obtained by the useless information(e.g.information header)via the nodes in the particular position sending information carrier periodically,thus to reduce the total overhead.The information broadcasting mechanism is robust that can withstand a certain degree of malicious attacks.And then the thesis come up with an adaptive parameter selection mechanism,DV(deviation value)is used to filter broadcasting information under the condition of large network load,it can adjust the bandwidth of the load according to the specific bandwidth load condition to control the amount of data which has been sent.The adaptive selection of the DV is based on the current network load condition and specific local situation.The details as follow:using the Q-learning reinforcement learning method,calculating the return value with the change of state,and conducting trial-and-error training according to the behavior/return value,then using the approximation of function and the principle of fuzzy logic approximation to find the optimal Q function,thereby chose optimal behavior at each step.Due to the node learn from the adjacent nodes,so the introduction of additional overhead is very limited.Comparing the current corresponding aggregation algorithms in the vehicular network and analyzing the performance of latency,throughput,packet loss rate and network load,we conclude that MLAIA algorithm is more superior in terms of the overall performance.