Research On Digital Modulation Signals Recognition Based On Wavelet Transform

Modulation type plays an important role in distinguishing different modulated signals. The modulation type of a signal and its parameters must be acquired if we want to grasp its contents. Given a section of signal, the main objective of modulation recognition is to decide its modulation type and estimates its corresponding modulation parameters without priori knowledge about the information contents. With the rapid development of communication technology, the systems and modulation manners of communication signals become more and more complicated and various, and the circumstance of signals becomes increasing denseness, recognition difficulty more and more greatness.This paper studies the digital modulation recognition of several kinds of common communication signals with the tools of wavelet transform. Using Mallat algorithm and Multi-Resolution Analysis, this paper decomposes signals in multi-level and noise reduction, constitutes eigenvector with the power of per-level details with wavelet packet, and proves that it is insensitive to noise. Improving recognition the ability of digital modulation signals in the case of low SNR(0dB-15dB) by choosing appropriate wavelet. Classifier are an important process in the modulation recognition. BP learning algorithm with momentum can avoid constringency local minimum points. In the end, this article devises a new kind of wavelet neural networks combinatorial classifier. This classifier can distinguish diversiform digital modulation signals, such as OQPSK,16QAM,π/4 - QPSK. It could be proved that this classifier has steady success rate, in the condition of low SNR (0dB-15dB) by the computer simulation experiments.