Feature Extraction And Modulation Identification Of Underwater Acoustic Communication Signal

The modulation identification of underwater acoustic communication signals under noncooperative conditions is an important topic.Under non-cooperative conditions,the wide range of signals' parameters,marine ambient noise,multipath effects,and Doppler frequency shifts are all obstacles to the identification.This paper studies the methods of identifying four underwater acoustic communication signals(BPSK,QPSK,2FSK,4FSK)under non-cooperative conditions.(1)This paper analyzes the time-frequency characteristics of various underwater acoustic communication signals in detail and complete the following research work:(1)For the wide range of parameters,the analysis revealed obvious limitations of traditional frequency analysis methods.Wavelet transform with multi-scale characteristics has better adaptability and analysis capability for unknown signals with wide frequency range.(2)In this paper,an adaptive mother-wavelet-selection method with local parameter pre-estimation and a non-cooperative modulation identification method are proposed.For different frequency intervals and bandwidth ranges,the analysis found differences in the analytical ability of the mother wavelets with different parameters.The mother wavelet parameters are selected by pre-estimating the parameters of the received signal to achieve adaptive optimization.(3)For the high-dimensional time-frequency analysis results,this paper extracts the time-frequency features using the fuzzy c-mean clustering algorithm in order to reduce dimensionality,and uses the Kneedle algorithm to improve the discrimination of small-frequency-deviation FSK signals.(4)At the same time,this paper also analyzes in detail the effects of special phenomena such as multipath effects and Doppler frequency shifts on time-frequency features.(2)In response to the problem that time-frequency features cannot identify BPSK and QPSK,this paper analyzes and uses cyclic spectrum feature to identify,and completes the following research works:(1)In this paper,the cyclic spectrum differences between PSK signals are elaborated from both theoretical and image analysis perspectives,and an identification feature is proposed based on them.(2)A local selective algorithm based on the FAM algorithm is proposed for the complex problem of cyclic spectrum computation.This algorithm,combined with the FAM algorithm,selectively computes CSA cells based on pre-estimated parameters,improving the computational efficiency and speed of feature extraction.(3)At the same time,this paper also analyzes in detail the effects of special phenomena such as ocean multipath effects and Doppler frequency shifts on cyclic spectrum features.(3)Based on the core processor OMAPL138 with dual-core architecture and unmanned platform requirements,a dual-core parallel software framework for multi-feature extraction is proposed.Control,management and computation are allocated to different processors through a modular functional configuration that combines the respective capability characteristics of the two processors,improving the resource utilization of the processors.An adaptive pulse filtering method is proposed,which can dynamically change the effective pulse threshold according to changes in the marine environment,filter the environmental noise,improve the system autonomy and reduce the energy consumption of the system operation.In summary,the unmanned platform implements a low-power,autonomous,non-cooperative modulation identification of underwater acoustic communication signals,and the simulation data and sea trial data are tested using the modulation identification test system,and the results validate the effectiveness of this method.