Research On Parallel Compression Technology Of Real-time Data

In recent years, with a high development of the automation and intelligent in the fieldof industrial control, the capacity to process and storage industrial real-time data is furtherimproving. In order to relieve the pressure in data processing and storage, DataCompression technology is necessary in the processing industrial real-time data.Compared with the traditional industrial control, the amount of data is much hugerand the interaction between data is more and more frequent in the automation of industrialcontrol. This requires the data compression technology should have greater real-timeperformance, higher data concurrency and better adaptability for different situations of theindustrial real-time data. On the one hand it means that data compression technologyshould be charge with more business data, on the other hand it also require that datacompression should not only not affect the real-time performance of the whole system, butalso be improved the compression concurrency as far as possible. This sets the newrequirements and challenges to the industrial real-time data compression technology.With the characteristics of the real-time data in the automation of industrial control,the paper designs a whole parallel adaptive compression solution of real-time data. Thissolution improved correlative compression algorithm, at the same time, it also based onthe parallel computational model which is called Compute Unified Device Architecture.Inthis model, GPU takes responsibility for the extra computation in compressing anddecompressing to solve the shortage of the real-time performance, concurrency andadaptability in processing the massive real-time data in the automation of industrialcontrol. The whole solution is modularized and is divided into three modules. Thelossy-compression module provides compression service for the real-time data stream, andit is responsible for the concurrency and adaptability in compressing real-time data. Thelossy-decompression module provides interpolation service in decompressing, and itensures the accuracy and efficiency of the decompression. The two-module system is aloosely coupled architecture, each responsible for separate functions and optimization.At last, this paper has a test and analysis for the improved compression algorithm andinterpolation algorithm. The result shows that the performance of these two algorithms israised effectively after using GPU to execute the parallel computing. It also verifies thatthe parallel compression solution which is designed in this paper is better for the industrialreal-time data in the automation of industrial control than before.