Realization Of Communication Signal Modulation Type Recognition Algorithm

Modulation type recognition can provide parameters basis for signal demodulation,which is a very important link in non cooperative communication.This paper studies the modulation type recognition algorithm of communication signal.Firstly,the basic research situation of communication signal modulation type recognition algorithms is briefly introduced.According to the needs of scientific research projects,the modulation type recognition algorithm based on feature extraction is studied,which has good real-time performance and is widely used.A decision tree based recognition algorithm is studied.Seven modulation signals to be identify,including AM,FM,BFSK,QPSK,16 QAM,DSSS and FHSS,are analyzed.Six of high-order cumulant feature parameters and instantaneous feature parameters are selected to identify modulation types.The algorithm has good recognition performance and good real-time performance.In order to further improve the recognition performance under the condition of low signal-to-noise ratio,a random forest classifier is used,and three of the feature parameters are selected by the error rate method of out of bag data based on six feature parameters.The simulation results of the two algorithms show that the recognition performance of random forest based recognition algorithm is better than that of decision tree based recognition algorithm,but the recognition algorithm based on decision tree has low delay and is easy to implement.Then,the hardware and software of modulation type recognition system are designed and implemented.The core modules of the system hardware platform mainly include AD9361,FPGA,DDR3 and DSP.AD9361 realizes RF signal receiving,analog-to-digital conversion and down conversion,FPGA realizes zero IF signal receiving and transmission,DDR3 provides high-speed storage area,and DSP mainly completes signal modulation type identification and system control.The software parts of the system mainly include: the zero IF signal transmission and communication between DSP and FPGA,and the realization of modulation type recognition algorithm based on TMS320C6678.The system can recognize eight signals at the same time with high real-time,high concurrency and high accuracy.Finally,the modulation type recognition system is tested.The RF signal is preprocessed by AD9361 to get zero IF signal,which is transmitted to DSP by FPGA and DDR3 for FFT.The accuracy of the signal spectrum is tested and the effectiveness of the whole system is verified.The real-time performance,recognition performance and multi signals recognition of the recognition algorithms based on decision tree and random forest are tested.The test results show that the system can recognize eight communication signals in real time,and the recognition performance is good.The recognition performance of the random forest based recognition algorithm is better than that of the decision tree based.The recognition algorithm based on decision tree has better real-time performance and simpler structure,and is easy to implement.It has been applied in the national defense research project.