Modulation Recognition And Estimation Of Communication Signal Based On Spectral Analysis

Modulation recognition and estimation of communication signal is important for electronic warfare and spectrum policing. There is no mature project to solve the problem of modulation recognition, because of the exist of contradiction between anti-noise performance and complexity. An algorithm which is simple and of good recognition performance is the mainstream of the research. This paper focus on spectral analysis on signal used for the modulation recognition and estimation of communication signal, because spectral analysis have some advantages as opposed to algorithm based on transient characteristics and higher order theory. The aim is to realize modulation recognition and estimation of communication signal with a simplest method.The signal refered to in the paper is:2ASK,4ASK,2FSK,4FSK,BPSK,QPSK,OQPSK and PI/4QPSK.The paper analyse all kinds of modulation signals, including square spectrum, the fourth power spectrum and cyclic spectral density, and sum up a set of characteristic parameter based on spectrum, and accomplish a method of modulation recognition which is validated according to mathematical reasoning and problem solving and simulation. The performance of characteristic parameter under variable snr is simulated and is validated for modulation recognition. Considering real-timing of algorithm, FFT is used for estimation of signal spectrum, and the specific steps are expressed. Then the paper makes a comparation with DMRAs in the context of anti-noise performance, area of recognition and algorithm complexity to approve the superiority of spectral analysis.In addition, this paper accomplishes the estimation of carrier frequency and chip rate for six kinds of basic modulation signals:2ASK,4ASK,2FSK,4FSK, BPSK, QPSK. For the ASK signals of which the carrier frequency is not restrained, searthing the peak value in its spectrum can estimate the carrier frequency, and its anti-noise performance is very good, then simulation is done under the variable snr to verify the recognition rate. for the BPSK/QPSK signals, carrier frequency is estimated base on the fact that square spectrum of BPSK and The fourth power spectrum of QPSK exists spectral line which signs the carrier frequency. Aim at the problem of noise power increasing in power spectrum, algorithm is made work better with the method of accumulating. It makes clear that the requirment of snr descend 8 dB. for estimation, based on the fact that instantaneous parameters of ASK,FSK is in accord with its base signal, and using the temporal difference extract the information of base signal jump instead of haar wavelet transform. Then the simulation is done under the variable snr to verify the recognition rate. For BPSK/QPSK, the spectral line in spectum of jump data signs the chip rate. Algorithm is made work better. It makes clear that the requirement of SNR descends 5 dB and requires less operand. Finally, estimation based on cyclic spectral is discussed and researthed.