On improving the accuracy of a wavelet based identifier to classify CDMA signal and GSM signal

Modulation identification has many applications, such as interference detection in wireless communication systems. This paper studies the use of wavelet transform to distinguish between CDMA signal and GSM signal. The modulation types of these two signals are different. The former has QPSK modulation while the latter has GMSK modulation. The approach is to use wavelet transform to extract the different transient behaviors of the two signals resulted from their modulation types, and apply template matching in wavelet transform domain for identification. Previous work only applies the wavelet transform at a single scale to an input signal. This paper improves the accuracy of the identifier by combining the identification results at different scales. When E/sub b//N/sub 0/ is -13 dB and the observation period is about 0.5 ms, simulation shows that the percentage of correct identification is 100% for GSM signal and 97.5% for CDMA signal.