Research On Modulation Recognition Technology Based On Software-Defined Radio

Modulation recognition is a key technology in non-cooperative communication scenarios such as electronic reconnaissance,electronic countermeasures,spectrum monitoring,and is widely used in military and civil fields.With the rapid development of modern communication technology,the signal types are more and more abundant,and the electromagnetic environment is more and more complex,which not only makes the modulation recognition more and more difficult,but also makes the traditional hardware circuit more and more difficult to meet the current diverse needs.Software radio provides support for modulation recognition engineering applications because of its good versatility,real-time performance and portability.In this paper,the key technologies of modulation recognition are studied in the context of non-cooperative communication.Aiming at the automatic modulation recognition of communication signals,this paper uses USRP to build a modulation recognition system of communication signals.The system functions mainly include signal detection,parameter estimation and modulation recognition.The research work of this paper is as follows:(1)The signal detection is realized by power spectrum analysis and maximum minimum average ratio of Welch spectrum estimation,and the detection performance of the two algorithms under different signal-to-noise ratios is compared and analyzed by simulation experiments.In the carrier frequency estimation,aiming at the inherent systematic error of the frequency domain center method and the low estimation accuracy of MFSK signal,this paper proposes the cyclic peak center of gravity and the cyclic first frequency point algorithm.The algorithm can effectively avoid the center of gravity offset by cyclic calculation of the local power spectrum.The simulation results show that,When SNR≥-5d B,the relative error rate of carrier frequency estimation of each signal is below 4%.(2)Aiming at the problems of limited recognition signal types,poor recognition of high-order modulation and great influence of noise in traditional instantaneous feature modulation recognition algorithm,this paper optimizes from three aspects: wavelet transform low frequency reconstruction denoising of basic instantaneous information,construction of new feature parameters and design of classifier,which expands the set of identifiable signals and improves the recognition accuracy.The simulation results show that,When the SNR is as low as 0d B,the average recognition rate of the improved algorithm for each signal can reach more than 91.2%.In the symbol rate estimation,wavelet transform is used to reconstruct the instantaneous information sequence with high frequency to realize the symbol rate estimation,and the effectiveness of the proposed method is verified by simulation experiments.(3)The system experiment platform is built and tested,and the USRP-2920 is used to receive and send real signals to verify the related algorithms studied in this paper.The measured results show that the parameter estimation errors of 4ASK,4FSK,4PSK and 16 QAM signals are all below 0.32%,and the average modulation recognition accuracy is above 98% when SNR≥15d B,which verifies the engineering practicability of the proposed algorithm.