Testbed and experimental analysis of automatic modulation classifier for non-uniformly sampled signal

In dynamic spectrum environment, radio terminals adapt the modulation type of transmission according to channel parameters such as noise, fading etc. to meet the desired quality of service. Subsequently, receiver needs automatic modulation classifier (AMC) to identify the modulation type of transmitted signal if not informed explicitly. Also, an opportunistic spectrum access based cognitive radio needs AMC to accurately characterize the status of frequency bands. In this paper, USRP testbed has been developed to analyze the performance of cumulant based AMC in real radio environment for BPSK, QPSK, 16-QAM and 64-QAM modulation schemes. In the proposed testbed, the received signal is either uniformly sampled at Nyquist rate or non-uniformly sampled at sub-Nyquist rate and subsequently reconstructed. For non-uniform sampling (NUS), multi-coset sampling and orthogonal matching pursuit based reconstruction approach is realized. Experimental results, for various modulation types, antenna gains and various distances between transmitter and receiver, show that performance of AMC does not degrade significantly with NUS given that received signal is approximately sparse in frequency.