Research On Algorithm Of Recognition For Multi-modulation Broadband Communication Signals Based On Cyclostationarity

The modulation type of communication signals provides an important basis for receiving and processing communication signals and the extraction of the information carried by signals.In the field of cognitive radio,vehicle communication,space electromagnetic spectrum monitoring,military communications and electronic confrontation and other applications,the detection and recognition of an unknown received signal play a pivotal role.In addition to the basic contents of the signal detection,multi-domain analysis and parameter estimation,the modulation scheme of an unknown signal must be identified by modulation classification.On the basis of which the corresponding demodulation algorithm is loaded to complete the demodulation and the transmitted information extraction,and then other tasks can be performed.Therefore,modulation recognition is considered as an indispensable but sophisticated mechanism for the unknown information detection and acquisition.In this thesis,combined with a new signal processing theory,named digital signal processing on graphs(DSP_G),a novel modulation classification method for broadband communication signals is proposed.The proposed modulation recognition scheme is based on the cyclostationary theory and DSP_G theory.The main contents are as following:1.Based on the features of the cycle spectrum,the modulated signal is first transformed from the cycle-frequency domain into the graph domain.The basis for the graph presentation of the cyclic spectrum is each cycle frequency of the cyclic spectrum matrix.2.According to the graph presentation on the cyclic-spectrum domain,a corresponding method of feature extraction and a new modulation classifier are proposed.Firstly,the sparse adjacency matrices can be easily constructed from the training data and the test data.Then the Hamming distance is employed to design the modulation classifier.3.Since the MPSK / MQAM signal can not be recognized based on the cyclic spectrum,the recognition scheme of MPSK / MQAM signals based on fourth-order cyclic cumulants and DSP_G theory is proposed.In order to get the concise graph-presentation,the transformation from fourth-order cyclic cumulants to the two-dimensional matrix is essential.Finally,simulation results of modulation recognition algorithm based on the cyclostationary theory and DSP_G theory are proposed in this thesis.At the same time,the algorithm is compared with the existing modulation-classification method,and the effects of frequency offset and timing error on the performance of the proposed algorithm are analyzed.The algorithm complexity of the proposed algorithm and the existing algorithm is compared.Simulation results demonstrate the greatly enhanced performance of the proposed algorithm.Although the recognition performance is affected by frequency offset and timing error,the time complexity of this algorithm is low.On the basis of this,simulation results of MPSK / MQAM signal recognition algorithm based on fourth-order cyclic cumulants and DSP_G theory are provided.The simulation results show that performance of the 8PSK signal recognition is very good in the identification of MPSK / MQAM signals.