The Research And Application Of Modern Power Spectrum Estimate In Radio Interference Signal Detection

Spectrum resource is a completely scarce one, within the limited frequency bandwidth, the administrative department of the radio has already programmed the frequency which be employed by every radio services. While, it’s commonly to run into the interfering signal as the radio traffic services are normally used, which has significant impact on the radio traffic services, among these interfering services, especially the interference of illegal radio station are the most serious one. In order to exclude all the interfering signals to make sure the normal and safe operation of the radio traffic services, spectrum resource should have been monitored. With the openness of radio transmission and the complexity during the transmission, it’s commonly occurred during the spectrum monitoring that the frequency point of the interfering signal is very close to the interfered one, or the less-severe interfering signal, these make it much harder to record the evidence of the radio spectrum monitor.The receiver used during the frequency spectrum monitoring and the function of upper-computer software determine the quality of monitoring. At present, most algorithm among all spectrum estimation algorithm took in upper-computer software are the classic ones. The classic algorithm has wonderful speed, the process of calculation is simple, but it is limited by the original signal on the minimum resolution. At the same time, because of cutting out the original data, it is inevitable to use the rectangle window, which has some negative impact on the resolution of the final signal. When we use the modern power spectrum estimation algorithm, the final resolution may not be determined by the original signal, thanks to some strategies have taken such as extrapolation, it’s possible to obtain better distinguishing ability and precision with the same length of the signal.The article based on software and hardware, which initially researched on the stationary signal and the statistic character of signal, and focused on some methods of spectrum estimation afterwards, a revised statistic spectral method with multi-frame has been introduced at the same time. Then we made plenty of simulation experiment of many methods on the platform of LabWindows/CVI and MATLAB. At last, we migrate the algorithm on the stage we do have, and we tested it in a realistic environment.We could get it according to the simulation experiments, it could get much higher resolution when taking the method of AR for estimating. But the background noise amplitude of the original signal shouldn’t be so large, or it would comes to false peaks. While, when we use the method of AR would take so much time. It’s unpractical to realize real-time monitoring in a practical application. However, when we take the method of Welch and the revised statistic spectral method with multi-frame, the effect doesn’t work as well as the method of AR. But these two methods could work with the low requirements of the original signal, otherwise, it has quick calculating speed, which is a spectrum estimation algorithm on load balance.After the migration of the algorithm to the stage of the software and the hardware we do have, we can show by the tests in the realistic environment, when we take the both methods of the revised statistic spectral method with multi-frame and Welch, it could enhance the variance of spectrum features and the resolution effectively, but we should pay attention that when the span of the spectrum comes to very narrow, it can often get a much better resolution effect to a micro signal. Meanwhile when we use the revised statistic spectral method with multi-frame, we could get a better effect when it works on the processing in frequency domain rather than the temporal processing.