Analysis And Application Of Stream Data Based On Infosphere Streams

At present, a new data schema named stream data appears in many applicationfields. Unlike traditional static data, stream data has features of constancy, massiveand time-critical. So it can’t meet the demand of real-time to processing stream datawith tradition method because of high-latency. And it is unuseful to introducedistributed-computing into processing stream data. So a new method is needed tosupport processing stream data, it is stream computing.Now the traditional method is widely used for processing stream data inIntelligent Transportation System, and it can’t be able to satisfy the increasingdemands. Therefore, a plan has been put forward to processing massive, real-timestream data of ITS with stream computing and stream computing platform in thisthesis, in order to satisfying the increasing demand for real time by improvingprocessing speed and reducing processing delay. After in-depth analysis of streamcomputing and InfoSphere Streams, the real-time traffic monitoring system has beendesigned and finished.To evaluate the actual performance of the new solution, three applications withsame function have been developed, the first is based on traditional method, thesecond is based on stream computing but with single assembly line and the last isbased on stream computing with multi-assembly lines. The results of threeapplications have been analyzed. Considering the key performance quota that streamcomputing concerns, the specific criteria and experiment have been designed. Byexecuting the experiments and analyzing the result, it shows that stream computinghas more advantages on processing stream data, and the processing speed can bepromoted several times with multi-assembly lines, the feasibility and rationality ofintroducing stream computing into processing stream data in ITS is confirmed.