Research On The Algorithms Of Spectrum Sensing And Signal Modulation Recognitions

With the rapid development of wireless communication technology, the shortage of spectrum resources is an urgent problem to be solved, the cognitive radio technology can meet this requirement. As the key technology of cognitive radio to detect whether the authorized users exist or not, Spectrum perception plays an important role. At the same time, the cognitive system should have its own cognitive ability and the reallocation function. Realizing the recognition of authorized users’ signal modulation type can provide reliable reconstruction parameters for reconfiguration of cognitive radio, thus avoiding the disturbance of normal communication from cognitive users. Therefore, the spectrum sensing and signal modulation mode recognition have been studied in detail in this work.Firstly, the energy detection and cyclostationary detection algorithms commonly used in spectrum sensing at present are studied, and the influence of the size and uncertainty of the noise on the detection performance are discussed. Secondly, the double threshold detection algorithm based on energy and cyclostationary detection and the fouth-order cumulants algorithm is studied. Contrastive analysis of the energy detection, cyclic spectrum, double threshold detection and the fouth-order cumulants algorithm are given by simulation results. Under the premise of the existence of perceived signal, this paper identify the modulation mode of the detected authorized user. A joint method based on the high-order cumulants and cyclic spectrum with intelligent decision algorithm (neural network) is proposed to recognize the modulation schemes for digital signals in this thesis. A new featured parameter is extracted from the fourth-order and sixth-order cumulants of the digital signals. The semi-physical simulation and mixed programming of LabVIEW and MATLAB are used to validate the proposed algorithms. Simulation results show that the algorithms can recognize modulations{OFDM, BPSK, QPSK,2ASK,2FSK,4FSK, MSK, 16QAM,64QAM} with small signal-to-noise ratio (SNR). The average recognition rate is more than 94% with SNR greater or equal than 5 dB, which validates the effectiveness of the algorithms proposed in this work.