首页 | 本学科首页   官方微博 | 高级检索  
     检索      


Digital processing of radar signals using industrial computers with vector AltiVec units
Authors:Wojciech Ros?oniec
Institution:Faculty of Electronics and Information Technology, Warsaw University of Technology, Nowowiejska 15/19, 00-665 Warsaw, Poland
Abstract:Radar systems usually had used digital signal processors for signal processing in the past. Such an approach has changed after introduction of a new type of general purpose PowerPC processor with very fast vector units called AltiVec. This new type of processor could realize various tasks performed earlier by several specialized processors. For instance it could be used for digital signal processing, tracking, fusion of data or communication with other functional blocks of radar. Its versatility and speed proved to be a superior solution in modern radar systems. It could also be programmed using C language instead of an assembler, what facilitated software development. Unfortunately the memory subsystem of computers built using this type of processor appeared to be too slow, and consequently slowed down the calculations. Therefore, the signal-processing software written for PowerPC processors with AltiVec vector units had to be accordingly optimized. The paper presents various optimization techniques and their effect on mean processing time of signal-processing software.The computer system built of industrial computers connected by the internal Ethernet is also presented in the paper. Individual computers of this system contain the multi-core PowerPC processors equipped with AltiVec vector units. The experimental model of the system is used for real-time digital processing of the multi-stream radar signals. The presented results of theoretical and experimental investigations show that the system is an effective, universal and cheaper computational platform than the corresponding, traditional multiprocessor platforms using the signal processors. Selected example recipes of writing the optimum and reliable application software are also given.
Keywords:
本文献已被 ScienceDirect 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号