Design And Realization Of A Method Of Automatic Modulation Recognition Of OFDM

The automatic modulation identification as the medium step of the signal interception and demodulation is the method of feature extraction of the unknown signal, and then recognize the modulation through appropriate classification algorithm. With the development of modern communication technology rapidly, the signal modulation recognition were gradually concerned by more and more people, especially in the field of software radio, electronic countermeasures and spectrum management. Based on the previous studies, a set of characteristic parameters extracted from the frequency domain, high order cumulant domain and time domain, by support vector machine classifier, the automatically classification modulation of OFDM and single carrier signal is studied and improved in this thesis.Power spectrum shape of different modulation signal is not same, so the power spectrum can partly reflect the characteristics of modulation signals. Compared with single carrier signals, the power spectrum of OFDM has no obvious impact component and is relatively flat in-band spectrum, and the characteristics of the side lobe is relatively narrow. Based on power spectrum characteristic, three parameters used for OFDM signal and the signals of single carrier modulation recognition put forward, including 2ASK,16QAM, 2PSK,4PSK,8PSK, OQPSK, MSK,2FSK and 4FSK. In order to improve the applicability of the method, in this paper, power spectrum is normalized firstly. At last, the performances of various characteristic parameters are simulated.Higher-order cumulant has a value of zero when Gaussian white noise is greater than the second order, so it performs good noise immunity. A recognition method of OFDM and single carrier signals based on higher-order cumulant put forward in this paper, in which higher-order cumulant can be applied to the intermediate frequency signal but not only to baseband signal. The influence on signal amplitude of forming filter is eliminated by envelope mean normalization, and the processing is improved by intrinsic time-scale decomposition transforming signal, then the instantaneous amplitude module value of complex signals in a rotating component is extracted, after that normalized fourth-order cumulant of the module value as recognition parameters identifies OFDM, according to 5 kinds of improved and imposed characteristic parameters, eventually classification of 7 kinds of single carrier signals is completed, including 2ASK,16QAM,2PSK,4PSK,8PSK, 2FSK and 4FSK. Simulation results shows that when the SNR is greater than 9 dB, and the accuracy of recognition is greater than 90%.Finally, a demonstration system based on DSP and FPGA is introduced in the paper. The cache and interface components is realized by FPGA, and the algorithm is realized by DSP in the system, then the results conveyed to the upper computer on PC to display. In this paper, the performance of the algorithm on the hardware system was tested. The hardware system further proves the effectiveness and feasibility of the proposed algorithm in this paper.