Research And Implementation Of Mobile Sensing Oriented Data Processing Framework

Maturity of the mobile Internet and widespread popularization of intelligent terminals equipped with various sensors spawned a whole new field of study-mobile sensing. Mobile sensing takes mobile devices equipped with sensing ability as the carrier and takes people who carry sensing devices and their surroundings as sensing target. It is characterized by fully taking advantage of mobile devices such as smart phones which are carried by side to sense the people as well as the environment surrounded continuously, get the sensing data continuously, real-timely process and analyze the data, and thus obtain various behavior and phenomena like traffic congestion state and disease spreading trend, etc. in order to better provide people with a variety of services. How to process large-scale mobile sensing data real-timely and efficiently is one of the key problems to research and solve in this field.Focus on the above-mentioned problems, paper in-depth analyzed the concept, characteristics and related technologies of mobile sensing, thus focus on the research of stream-based large-scale mobile sensing data processing model, further more designed topology-based real-time map matching algorithm and stream-based fast traffic condition aggregation algorithm considering the characteristics of traffic sensing data on the background of intelligent traffic application, on this basis, did the system design and implement and carried out experimental verification. The main work is reflected in the following three aspects:Firstly, for large-scale mobile sensing data’s huge mount and real-time processing problem, proposed a new stream-based mobile sensing data processing model SSPM, the model is able to reasonably divide and merge mobile sensing data stream, support sensing data stream reorder, rule-based concurrent data transfer and high efficiently sensing data stream aggregation, to balance the scale and real-time characteristics of both.Secondly, on the background of intelligent traffic, for how to fast mapping and matching large-scale traffic sensing data stream, used an efficient method of zoning, proposed a topology-based real-time map matching algorithm which is capable of mapping and matching large-scale traffic sensing data stream; for how to achieve fast traffic condition aggregation, proposed a road segment average velocity based traffic condition aggregation algorithm which can efficiently achieve fast trafficcondition aggregation.Finally, did the system design and implementation on the basis of streaming processing platform Yahoo! S4, then set up an experimental environment and took massive real traffic sensing data stream as experimental data to test and verify the result of this study. In the end, took compare the comparing experiment with distributed data processing platform Hadoop, which showed that SSPM-based real-time traffic condition aggregation algorithm could aggregate the traffic condition more efficiently.