Growd Sensing User Location Prediction Model Based On GRU Network

As an era of rapid development of science and technology,the corresponding mobile terminal equipment has been more perfect and developed.Based on the current integrated circuit development has achieved effective results on the basis of mobile devices abandon the past single mode,its development is more diversified,its function has also changed from the initial communication into a large number of data fast processing.On this basis,Mobile Crowd Sensing(MCS),as a new Sensing mode application,has been warmly discussed by the majority of scholars.It has almost the same overall framework and implementation process as the popular Internet of Things mode in the past,and includes Mobile Sensing and crowdsourcing technology.With the further attention and in-depth understanding of swarm intelligence sensing technology from all walks of life,its more value has been fully explored and effectively applied in all fields of social life.Most of the group of mental perception application task devices are almost all in our life held by intelligent mobile devices such as mobile phone,camera,so the capacity of this kind of data acquisition masses can be registered intellectual perception platform become involved in perception task group of candidates,and the work shall be carried out in accordance with the instructions to get the corresponding reward.For example,in urban life,there are all kinds of times every day that require us to know information.At this time,the smart phones carried by citizens can be used to collect information.At the same time,under the constraints of time and budget,the acquisition of effective data information is extremely important and complex for the choice of users who can complete the perceived task.However,there is often a phenomenon that when we select users according to task requirements,we find that the number of users in the task area is not enough to support the completion of the task,resulting in the failure to effectively collect perceptual task information.If accurate prediction of user location can be made,it can help the platform to quickly determine the most likely users in the task location,so as to avoid the failure of task information collection caused by sparse user distribution.Therefore,the prediction of the user's future location is very important for swarm intelligence sensing applications.This paper proposed a user location prediction model based on Gated Recurring Unit(GRU).Firstly,the feasibility of applying various neural network algorithms in user location prediction is analyzed and compared.Combining with the actual road network information,a user location prediction model based on GRU network is built according to the analysis of the characteristic data of user location time series.Mean Squared Error(MSE)was used as evaluation criteria to optimize the number of iterations and the size of batches in network model parameters.The Adam optimizer is selected according to the cross entropy cost function.Finally,the simulation and comparative analysis of the proposed model prove that the proposed model can effectively predict the user's future location accurately.