Key Technology And Theory Of Wireless Communication Signal Modulation Recognition

Modulation automatic recognition of wireless communication is the basis of the research area in software radio, cognitive radio and spectrum sensing etc. It has extensive application in military and civil use so it gains the extensive attention among the researchers. Particularly for the setting air defense zone in the recent year, it is also a challenging research topic in how to achieve continuous innovation based on the existing automated recognition technology method, and the improvement of recognition for the recognition and monitoring the external aviation aircraft signal, especially for the automatic identification of the signal modulation method. The key technology and theory about modulation recognition method and algorithm of wireless communication system are researched and the main achieved results of research are following:1. The research is conducted on the key technology and theory about the N-dimensional block orthogonal modulation, demodulation recognition method, the reestablishment and recognition of MIMO system. The research is based on ordinary orthogonal modulation recognition, achieved data block demodulation by extracting basic feature vector from the received signal samples to estimate the signal parameters, and achieved recognition of signal modulation through the matrix conversion, The simulations results under the conditions of an additive Gauss white noise channel show that the method has a good performance of recognition to signal of N-dimensional block orthogonal modulation and calculating complexity level can be relieved compared with applying the recognition of all received signal vector. In addition, the key technology and theory about reestablishment and recognition of MIMO system is conducted in the research. Radial basic function network, support vector regression is used to establish initialized structure of network based on radial basic function and set the parameter of initial network, the algorithm of anneal ing dynamical learning is adopted to practice for the system identification network in the paper. Among the practice, the algorithm of particle swarm optimization iterative is firstly adopted to select the combination of the best learning rate, which makes MIMO system recognized by the identification network. For the selected MIMO system which is waited to be recognized with two inputs and two outputs, the result of simulation shows that the performance of system identification is better than the frequently-use least square algorithm or the gradient descent algorithm in the current process of parameters optimization based on radial basic function network.2. The key technology and theory about MLP neural network communication modulation recognition method is studied. As there are the problems about slow speed, the emergence of false saturation phenomenon etc in signal recognition of multi-layer perceptron neural network classifier based on back-propagation algorithm existing, the combined feature module selected by algorithm of bee colony are used. Meanwhile, three different algorithms of quick prop, super schemes for adaptation of error back-propagation algorithm, conjugate gradient are presented used in multilayer perceptron neural network classifier to achieve the automatic recognition of communication signal in this paper and higher recognition rate compare with error back-propagation algorithm. To solve the problem that the recognition rate of the signal modulation recognition methods based on the clustering algorithm is low under the low SNR, the characteristic parameter of the signal is extracted by using a clustering algorithm, neural network is trained by using algorithms of variable gradient correction (Polak-Ribiere) so as to enhance the rate of convergence, improve performance of recognition under the low SNR and achieve modulation recognition of the signal based on the modulation system of the constellation diagram. The simulation result shows that the modulation recognition rate is obviously improved compared with the single application of clustering algorithm or neural network recognition based on BP algorithm under the low SNR.3. The key technology and theory about modulation recognition method of mixed modulation signal, single carrier and multicarrier digital modulation based on decision-making theory algorithm are researched. The tree classifier, the recognition steps which are suitable for mixed modulation signal, single carrier and multicarrier digital modulation are put forward respectively in this paper. A phase folding algorithm to correct the influence of phase folding was first used in instantaneous phase extraction and the accuracy of characteristic parameters was improved. The characteristic vector which are composition of the number of subcarrier signal, envelope variance of mean normalization and algorithm of the statistical value of subcarrier signal instantaneous amplitude distribution area are first used in recognition of outer modulation and inner modulation respectively so as to reduce the noise interference and improve the accuracy of characteristic parameters. The simulation result shows that this method can achieve better recognition performance compared with the existing mixed modulation, single carrier and multicarrier digital modulation recognition methods.4. The key technology and theory about VHF modulation classification recognition method based on the algorithm of first-order cyclic moment is researched. In view of the problem of modulation recognition algorithm existing about low recognition rate under environment condition of low SNR, the algorithm of first-order cyclic moment which are first presented is used in band signal of VHF modulation classification recognition in this paper and the recognition rate increased significantly. The first-order cycle frequencies at which estimated first-order cyclic moment magnitudes exceed the cut-off value were selected as candidate cycle frequencies based on estimation first-order cyclic moment at first-order cycle frequency series, then decided to choose the number of candidate cycle frequency by a cyclostationarity test to achieve signal modulation classification recognition. The simulation result shows that the recognition rate can achieve bigger improvement compared with existing modulation recognition algorithm under the low SNR.