Digital Modulated Signal Recognition Based On Intrinsic Time-scale Decomposition

Automatic recognition of modulation is the method of extracting feature parameters and then recognizing the modulation scheme by proper classification algorithm. With the rapid developing of communication technology and denser circumstance of signals, the system and modulation schemes of communication signals become more and more complicated and various. It leads to the hot research of modulation recognition.This thesis studies digital modulated signal recognition algorithms based on the intrinsic time-scale decomposition. The main contributions are as follows:1. Based on the intrinsic time-scale decomposition algorithm, the instantaneous amplitude, phase and frequency is precisely defined. This algorithm is simple and suited for the real-time operation. A decision-tree recognition algorithm based on the intrinsic time-scale decomposition is proposed. Simulation shows that when the improved threshold setting for digital modulated signals classification is used and the SNR is 10dB, the average recognition ratio in this algorithm can achieve 98.6%.2. Based on feature parameters'extraction from the time domain, the spectrum domain as well as the fractal domain, separability measures-ratio F and ratio D weigh the feature parameters. With in-class compactness and between-class separateness, three feature combinations are choosed for signal recognition.3. Based on the improved distance measure classification, a valid recognition is obtained only using three parameters in each feature combination. And the effect of different length samples on the recognition ratio is studied. Simulation shows that when SNR is 5dB, the average recognition ratio achieves 94%, which confirms the validity of separability measures.