Research And Implementation Of Adaptive Channelization Filter-banks And Automatic Modulation Classification

Digital receiver technology based on SDR(Software Defined Radio) architecture is widely used in the field of communication signal reception because of its high flexibility, reusability and other advantages.This paper focuses on the channelization filterbanks which belongs to the front-end technology of SDR, and automatic modulation recognition classified to the back-end of SDR. System implementation is completed according to the relevant algorithms on the platform of signal processing systems. The main work and achievements of this paper are summarized as follows:1. The basic theory of SDR reception is introduced in the beginning. Theoretical analysis, spectral analysis of output data and implementation of FPGA-based systems are made towards the technology of DDC(Digital Down Converter), the first stage of pre-processing in the reception of IF digital signal, as represented in tertiary structure of sampling rate conversion. The guidelines to be followed when designing actual multi-stage DDC are summarized in the end.2. Aiming at the dynamic blind reception of the uniform distribution IF signals, An adaptive channelization IP core is put forward based on FPGA. The number of subchannels, subchannel order and prototype filter coefficients are dynamically configurable. Design principles, its structure, Modelsim simulation results and on board test results are introduced and analyzed. Based on the IP core, a dynamic blind reception algorithm is proposed for the uniform distribution IF signals. By means of receiving signal without blind-zone, Spectrum splicing, parameter estimation and the receiver’s adaptive correction, the uniform distribution IF signals can be channelized received in the absence of priori information. The theoretical basis of each part is analysed and derived. Finally, the algorithm is simulated and confirmed on the signal processing platform.3. Aiming at the symbol rate estimation of MFSK signal, an algorithm is put forward based on the wavelet ridge feature extraction. By means of extracting wavelet ridge, filtering it, and clustering the distance of zero-crossing, the effect of Multi-Path and Doppler phenomenon in estimation of the symbol rate can be overcome, which is shown in the Simulation experiments. The theoretical basis of each part is analysed and derived.4. Based on above studies of wavelet ridge, a recognition algorithm is proposed used for the distinguishing between MFSK and Amplitude-phase modulated signal via multipath channel, which achieves high recognition rate in low SNR environment compared with traditional algorithms. Then another algorithm of automatic modulation classification based on high-order cumulant features is raised for the amplitude-phase modulated signal via multipath channel. By means of computing cumulants, construction of nonlinear equations and Gauss-Newton iterative method, it can effectively estimate the channel interference factor. Compared with fourth moment quotient algorithms, it reduces the impact of the distribution of channel factor on the recognition performance, and is no longer limited to a particular multipath channel, which shows stronger robustness.5. T he functional module construction of DDC, channe lization filterbanks, and automatic modulation recognition are completed on the signal processing platform. T he test results show that it can achieve correct sampling rate conversion, spectral shift and sub-channel filtering stably according to configuration parameters using DDC and channelization filterbanks module. Modulation recognition module also i dentifies the modulation style of the test signal. Anticipative results are achieved.