Research On Delay Optimization Schemes In Vehicular Sensor Cloud System

Vehicular sensor cloud system promotes the development of smart cities.Wireless sensor devices,vehicles,mobile edge computing servers,and cloud servers deployed in the vehicular sensor cloud system,these devices facilitate the collection,computing,storage,and sharing of data in current application scenarios.How to transmit data packets containing massive data to the control center with the lowest possible delay and ensure the efficient processing of the monitored events is very important for the related applications of the vehicular sensor cloud system.This thesis firstly proposes a Dynamic Waiting Timer-based Delay Optimization scheme(DWDT)to solve the problem of high transmission delay in the process of data collection due to fixed waiting timers for sensor nodes in wireless sensor networks under the vehicular sensor cloud system.This scheme changes the lack of a fixed waiting timer in the previous MAC protocol through a dynamic waiting timer adjustment algorithm.This algorithm dynamically adjusts the node’s waiting timer so that its value can always follow the optimal value,thereby reducing the data transmission delay.Based on this algorithm,the DWT-MAC protocol is proposed.Although the DWT-MAC protocol cannot guarantee that the value of the waiting time is always the optimal value,sensor nodes can always quickly adjust the value of their timer adaptively,making the wait timer trail the optimal value.The DWT-MAC protocol is very suitable for networks with dynamic data changes.A large number of experimental results show that the DWT-MAC protocol can effectively reduce the average transmission delay of high-priority packets and the total network transmission delay.The average transmission delay of the highest priority packets in the sensor network reaches 49.3%.DWDT can effectively reduce the total data processing delay of the vehicular sensor cloud system by quickly transmitting data to the sensor nodes.In addition,this thesis proposes an MDP-based Intelligence Data Processing Scheme(MIDP)to solve the problem of difficult decisionmaking of data processing objects which resulting in high data processing delay under the vehicular network in the vehicular sensor cloud system.The data processing problem in the vehicular network is described and modeled as a Markov Decision Process(MDP)under MIDP,and a corresponding MDP adjustment strategy is proposed for the problem that the model cannot be reused in the vehicular network with damaged servers,which ensures vehicles in vehicular network with server damage can still make intelligent decisions about data processing under MIDP.MIDP adopts the Asynchronous Advantage Actor-Critic(A3C)algorithm to solve the constructed MDP problem,and guides vehicles to complete intelligent decision-making through reinforcement learning,thereby reducing the data processing delay.A large number of experimental results show that,compared with the Expect Data Processing(EDP)scheme and the Immediately Data Processing(IDP)scheme,the MIDP scheme proposed in this thesis can reduce the data processing delay while ensuring a high data processing completion rate.The average delay of vehicle network data processing is reduced by 29.93% and 29.99% at the highest,while the completion rate of data processing is increased by 66.6%.This solution can further reduce the total data processing delay in the vehicular sensor cloud system through fast data processing.The two schemes proposed in this thesis focus on reducing the data transmission delay in the wireless sensor network under the vehicular sensor cloud system,and the data processing delay in the vehicular networks after sensor nodes transmit data packets to vehicles.Under the two proposed schemes,the total delay from collecting data from sensor nodes in the vehicular sensor cloud system to uploading data to servers for computation through the vehicle as a relay is greatly reduced.