an adaptive algorithm selection purposes
GPU parallel computing
The problem of allocation targets on a background noise in phased antenna arrays is very important today. Until now, the problem of noise suppression in the radar is not solved until the end. A few interference in the range of the radar remove it from the system. Adaptive algorithms can be easily suppress noise, while weakening the useful signal by only a few decibels. However, these algorithms are computationally intensive, tasks, and they have to work on radar stations in real time. The first adaptive algorithms have appeared two decades ago, but only in computer technology today has allowed to get close to being able to implement them in practice. Use special processor and supercomputers is expensive, they take up much space and require specially written software for them. The new models are constantly increasing number of processor execution cores. To use all the computing resources of modern multi-core processors, software running multiple threads for processing. The most powerful is by far the graphics processor, which contains up to 512 cores.
Using graphics processors, algorithms, data processing can operate in real time, which is necessary for detection purposes in phased antenna arrays.
This paper has been implemented adaptive algorithm selection purposes in the background noise in the radar with active phased array. The algorithm was tested on different types of graphics processors. The resulting time processing of signals from all directions, inspect the radar compared to the time spent on an overview of this space by the radar. The algorithm was tested on the computer system with three GPUs.