Simulation Analysis On Near-field Source Localization Under The Complicated Ocean Noise Environment

The complicated ocean physical characteristics lead to the uncertainty of theacoustic propagation channel including time change and air change, simultaneouslythe high-precision underwater target positioning is extensively applied in sonar, radar,underwater exploration as well as many other areas, which has been always a hot anddifficult research in the underwater acoustic field. Currently the research andapplication of underwater target localization technology mainly focused on thefar-field source localization, while the development of underwater robots also requiresclear demands of the near-field source high-precision localization technology. But,based on several ideal assumptions, it possesses high computational complexity,whose research and application lags far behind the far-field source localization. It is ofcrucial theoretical and practical significance to research the near-field sourcelocalization algorithms under actual complicated ocean environment and explore newhigh-precision near-field source localization technologies.Based on the summarization and analysis of the complicated marine environmentnoise characteristics, this paper concludes four complex ocean noise positioningmodels, simulates and analyzes the performance of existing near-field sourcelocalization methods, proposes two effective algorithms based on additive noise tocope with the mixed far-field and near-field source localization problem, and developsa set of simulation software simultaneously, which can verify the performance of tenalgorithms. The main work and innovations are as follows:(1) For the different noise environment of shallow water and deep sea, this paperanalyzes the origination of sailing noise, wind noise and other noise, then, generalizesthe sound source level, frequency distribution, stability and statistical distribution ofthe complicated additive noise and multiplicative noise, and concludes four realisticocean localization models, including the Gaussian noise localization model, thenon-uniform noise localization model, the impact noise localization model, and the coexistence model of multiplicative noise as well as additive noise.Based on the summarization of existing near-field source localization algorithms,this paper simulates and analyzes the RMSE performance of these methods under thecomplicated marine environment, including the Gaussian white noise, Gaussiancolored noise and non-uniform noise, then analyzes the parameter estimationperformance and evaluates the applicability where both multiplicative noise andadditive noise are considered simultaneously.(2) The DOAs and powers of far-field sources for the first agrithm can be estimatedfrom the MUSIC specture function. Then, an oblique projection technique is adoptedto eliminate the obtained far-field interference. Finally, both DOA and rangeestimations of near-field sources can be performed, exploiting the symmetry propertyof ULA geometry. Compared with the second-order MUSIC method, the resultantalgorithm provides an improved accuracy, as well as alleviates the aperture loss.(3) By exploiting the symmetric property of the divided subarrays, a rotationalfactor is firstly constructed for the second agrithm, in which the DOA information iscontained. Then, the DOAs estimations for both far-field and near-field sources can beperformed from a spectral search procedure. Finally, by substituting the obtainedazimuth back into the two-dimensional MUSIC spectral function, the rangeestimations for near-field sources can be detected. Compared with the previous works,the resultant algorithm has a low computational burden, provides the improvedaccuracy, as well as requires neither two-dimensional search nor parameter-pairingprocedure.(4) Referencing the signal processing of underwater target positioning system, wedevelop a simulation analysis software platform, which shows some applicability andfunction expansion easily. The simulation software can provide several functions,such as setting the parameters of each module, signals types selection, formationsselection, algorithms selection, the display of algorithm simulation results, and so on.Besides it can also supply additive noise, multiplicative noise, the coexistence modelof multiplicative noise as well as additive noise, ten localization algorithms, and tenarray formations, which can analyze the estimation accuracy of the near-field sourcelocalization algorithm.Above study is the exploration of the near-field source localization technologyunder the complicated marine noise environment, while it also provides a referencefor the research of array signal processing issues.