The Research And Simulation Implementation Of Composite Signal Recognition Based On FPGA

With the rapid development of communication technology, the communication environment increasingly complex, the aliasing phenomenon between various modulation signal is becoming more and more frequent. Therefore, as the pre-work of interference suppression technique and composite signal separation technology, the composite signal identification is particularly important both in military and civilian areas.Begin with MPSK and 16QAM, the signal source is divided into three categories, including single signal, symmetrical composite signal and asymmetric composite signal, a total of 15 kinds of composite signal model; then, the characteristics of composite signal’s full sampling sequence constellation diagram as well as composite signal’s weak signals’ timing sampling sequence are introduced; next,5 parameters of full sampling sequence constellation diagram are used to complete the recognition of symmetry composite signal and the recognition of modulation mode,2 parameters of composite signal’s weak signals’ timing sampling sequence are used to complete weak signal detection, the estimation method of weak signals’ number is discussed too; then algorithm optimization is done according to the FPGA and the algorithm transplantation is completed; finally, the hardware implementation is completed using the Xilinx Virtex5 development platform.In the process of study, firstly the Matlab simulation is done to verify the feasibility of recognition algorithm, then the ModelSim simulation is done to verify that the algorithm can be transplanted into the FPGA, finally cooperating with ChipScope software the test result of the hardware implementation is observed in order to prove the accuracy of the hardware implementation.A kind of composite signal recognition algorithm which is suitable for FPGA is put forward, it has small computing complexity, less resource consumption, and its recognition performance is more outstanding. Algorithm simulation, functional simulation and the test of hardware implementation can prove that the recognition algorithm can recognize all kinds of the signal model when the SNR is no less than 15dB, and it can also recognize most of the signal model when the SNR is slightly less than 15dB.