Abstract:
The concept of the software defined radio is leading to the development of ultimate radio systems that can tune to any frequency band and receive any type of modulation scheme over the large spectrum with most of the processing done in software. In order to realize such a universal radio, it is required to develop and implement effective and efficient algorithms that can successfully identify different modulation schemes in use today.
Modulation classifier designed in this project identifies each digital modulation scheme from a set of frequently used modulation schemes, which includes MPSK, MQAM and MFSK schemes. Feature based approach has been used for modulation classification process in which first the value of a particular feature is extracted. Based upon the value of the feature the digital modulation scheme is identified. Earlier designs of modulation classifiers mostly work at an SNR level of 10 dB and above. Careful selection of features based on amplitude, phase and frequency content of the signal improved the performance of the modulation classifier even at very low SNR. The design proposed for modulation classifier in this project successfully identifies digital modulation schemes up to SNR level of 5 dB.
