Research And Implementation Of Mobile Intelligent Terminal Data Preprocessing Methods For Crowd Sensing

The process called “Crowd Sensing” means using ordinary users of intelligent mobile devices(like mobile phones and pads)as the basic units of collaboration through the mobile Internet consciously or unconsciously to the distribute sensing task,collect and store sensor data,and complete the large-scale and complex social perception task.Because of the development till now,crowd sensing is having huge impact in intelligent transportation,environmental monitoring,health care and other aspects.Sensor data are gathered quickly,and consist of small data pieces.And when it comes to the mobile terminal,there are problems such as the lack of processing abilities of hardware and lack of space.The existing related research has failed to effectively solve the mass and various sensor data preprocessing optimization problems at the acquisition end in crowd sensing process,thus the efficiency of sensing data transmission and application is reduced,while affecting the crowd sensing application operation.This paper,according to the characteristics of various types of sensor data of the mobile intelligent terminal and the crowd sensing applications' needs of data processing,put forward the improved preprocessing method from the existing,so as to achieve better effect of preprocessing.The work done in this paper is as follows:(1)According to the characteristics of various types of sensors used in mobile data acquisition for crowd sensing applications,such as data structure,sampling frequency,the amount of data transmission,error sources and real-time transmission demand,This paper divides all sensors into several categories,then clear out the demand of preprocessing optimization of various sensing data in crowd sensing application,while considering with the actual needs of crowd sensing application.(2)Aiming at the problem of too much time consuming by the algorithm caused by the big amount of data,and the problem of redundant data records caused by high sampling frequency in the numerical sensor data acquisition in the existing sensing application,this paper presents an improved sliding average method,change the large number of continuous data into several reliable and representative data by the dynamic window to achieve data compression,and reduce the time complexity of the average method by reduce the amount of data that need to be processed.(3)Aiming at the problem that the algorithm is too time-consuming caused by lack of optimization when processing high pixel images in denoising preprocessing,An improved extremum median filtering method is proposed for the image data.The algorithm for detecting the extremum and median value in the filter window is optimized by locally sort the filter window matrix,and the gradient change of the filter window is used to accelerate the sorting of the moving window,which reduces the time complexity of the whole algorithm,while ensuring the effect of image denoising;(4)A transmission strategy optimization based on the “reference value and difference value" record is proposed,after grouping the data files,only the demarcation points of each group of data and the data points with large difference when compared with the demarcation points are recorded,which reduces the storage pressure and the amount of data transmission of mobile terminal,and improves the efficiency of data transmission.(5)The optimization methods mentioned above is implemented and validated to get the optimal parameters for these methods,and prove the efficiency of the optimization scheme proposed in this paper by comparing with the existing typical processing methods.