Crowd Sensing User Recruitment In Internet Of Vehicles Environment

The Internet of Vehicles provides strong support for crowd sensing.In the crowd sensing system in Io V,users(vehicles)can collect information from points of interest through movement and upload data to the cloud perception center.It can be applied to urban environment monitoring,real-time traffic monitoring and urban resource monitoring and so on.The network environment of Io V is integrated.So crowd sensing system in Io V can upload sensing data using cellular network or opportunistic network.The cellular network has high reliability but high cost,while the opportunistic network has low cost but low reliability.While the system needs reliable transmission of data,it also needs to consider the cost of recruiting users.According to data plans,users in Crowd sensing system in Io V can be divided into costsensitive users and cost-insensitive users,and cost-insensitive users have a smaller cost to upload data using the cellular network.Based on the historical trajectory information of vehicles,the historical candidate set of users corresponding to the perception region can be established.Bi LSTM model trained based on users’ historical trajectories.It can predict users’ position sequence in the future sensing period.Then their access probability distribution to the perception region in a certain time interval of the sensing period is obtained.With historical candidate set and probability distribution to the perception region,data transmission path can be constructed,and the connection probability of each path can be calculated.The user recruitment optimization problem is defined to select the data transmission path to minimize the cost while meeting the constraints of the coverage rate of interest points and the data delivery rate.This problem is an NP-hard problem.Two heuristic algorithms that give priority to the cost-effective path are used to solve the problem,select the appropriate path,and the users included in the path are eventually recruited.The singlehop heuristic recruitment algorithm restricts the data forwarding of the opportunistic path to one time to ensure a higher connection probability of the data path,and the multi-hop heuristic recruitment algorithm considers all data transmission paths.The performance evaluation is based on real data sets,and the results show that the optimized recruitment method can reduce the cost compared with the cellular network method.Compared with opportunistic networks,data delivery rates are higher and costs tend to approach opportunistic networks.Compared with the recruitment method based on the hidden Markov model,the Bi LSTM model used in the optimized recruitment method has a higher accuracy rate of position prediction,so the data delivery rate is higher and the perceived cost is lower.The single-hop heuristic recruitment algorithm limits the length of the path,reduces the complexity of the algorithm,and has a higher data delivery rate,but the perceived cost is higher than the multi-hop heuristic recruitment algorithm.The multihop heuristic recruitment algorithm considers the multi-hop transmission path,which further reduces the perception cost,but the complexity of constructing the multi-hop transmission path is higher.Specific performance indicators are also affected by the proportion of cost-sensitive users and cost-insensitive users.