The Design And Implementation Of Heterogeneous Data Fusion System Of Internet Of Things Based On Storm Engine

After years of research and development, multi sensor data fusion technology has achieved fruitful theoretical and applicative achievements. The progress of computer technology, communications technology, data processing technology and other related technology provides new impetus for the development of multi sensor data fusion technology. With the increase of sensor type and the expansion of the deployment area, there is an increasing number of application scenes of multi sensor data fusion technology. What’s more, modern data fusion algorithm is more complex, the sensing data structure is more diverse. All of above will limit the concurrency and scalability of the multi sensor data fusion system. The traditional centralized data fusion system is facing the severe performance challenge which can be resolved by increasing the parallel processing ability fundamentally.Recent years, there has been a great development in big data processing field. The core idea of the big data processing is that the complex operations are dispersed to multi processors to improve the processing ability by the method of distributed computing. Therefore, by combing the multi data fusion technology and the distributed computing technology, to design a real-time data fusion strategy with scalability and high parallel performance to serve data fusion algorithms will have important significance.This article designs a simple multi sensor data fusion system based on Storm computing engine. Storm is a kind of distributed computing engine which is similar to Hadoop MapReduce. The main difference of them is that Storm is a real-time streaming computing engine while Hadoop cannot do real-time work as an offline computing framework. The stream structure property of Storm fits to realize the function of all the modules of the data fusion system with algorithms. Since the multi sensor data fusion system based on Storm is implemented on the application layer to deploy every function modules on the Storm cluster environment to execute fusion algorithms and strategies, it has some differences between it and the traditional data fusion technology in theories and applications.The main research work of this article is:First, analyze the trend and problem of current data fusion technology, and putting forward the idea of increasing the parallel processing ability by distributed system. Then, investigate the properties of Strom engine which meet the performance and scalability requirements when the data fusion system connects a large number of sensors. Next, design the framework of the data fusion system based on Storm engine with the structure and function model of data fusion technology and narrate functions of all the modules. Last, deploy the system on the Storm cluster environment to validate that Storm can serve the fusion algorithms commendably.