| Summary: | 碩士 === 中華技術學院 === 電子工程研究所碩士班 === 94 === The electronic signal recognition system is composed of recognized processor and electronic signal database. Nowadays the electronic signal applications are extremely widespread. If I can accurately find the characteristic parameters of electronic signals, the recognizing rate will be improved apparently. And the help to the follow-up developments will be substantially.
Up to now, several algorithms for recognizing the interested electronic signals have been developed; all of them are written on MATLAB. The methodologies that algorithms employed include: the Euclidean distance in frequency-domain, the waveform cross-correlation in time-domain, the waveform cross-correlation in frequency-domain, the coherence function, the waveform of rising time, the instantaneous frequency of the rising-time waveform, the AR model spectral estimation, and the short-time Fourier transform (STFT). Results show that all of the algorithms have good performances, i.e., high classification rates. Besides in response to the real-time demand, the digital signal processor has been taken into account such as TMS320C6701(TI), to achieve high-speed analysis and high recognition of the intercepted signals.
The difficulties to implement the algorithms into hardware (DSP chip) lie in the transformation of languages and inference logic between MATLAB and DSP processor. To speed up the transfer of algorithms, the specific software suite, Code Composer Studio (CCS) of Texas Instruments (TI) integrated development software, has been employed to reconstruct the recognition system.
|
|---|
