以合成口徑演進週期頻譜法執行訊號源之定位

Abstract

本篇論文提出一套應用拖曳式聲納之水下訊號源定位與追蹤方法。吾人所 提之方法為演進週期頻譜法和資料-適應性演進法對非穩態性弦波信號頻 譜估計之延伸。首先，藉由合成口徑技巧，吾人採用單一聲納拖曳出一有 效口徑使之適用於演進週期頻譜法的架構來追蹤窄頻訊號源之方位。此架 構經由寬頻聚焦技巧可對寬頻訊號做處理來獲得更佳的時間-頻寬效果。 其次，為了補救低信號雜訊比對估計角度的影響，吾人將單一聲納擴展成 多個聲納連結的陣列結構，並提出一些建構在原先演進週期頻譜法和資 料-適應性演進法上的改良式波束合成器來增加系統之解析度。論文中亦 推導出一遞迥式來有效更新訊號源之方位估計。最後，吾人以電腦模擬來 驗証所提之方法在低信號雜訊比和出現鄰近訊號源的情況下，其仍能達到 不錯的追蹤效果。 In this thesis, a spatial spectrum estimation method is presented for sourcelocalization in an underwater environment using towed sensors. Conceptually,the method is developed as a modified version of the evolutionary periodogram(EP) and data- adaptive evolutionary (DAE) spectral estimators, which deal withthe tracking of time-varying sinusoidal signals. In particular, we start with the case of a single sensor and narrowband sources in which an effective aperture is formed using the synthetic aperture technique to acquire thespatial information of the sources. The method is then extended to the wideband case with the incorporation of focussing technique to obtain enhancedtime-bandwidth performance. The towed sensor can be augmented into an array ofmultiple sensors to form a towed array. With a beamformer attached to the array, the SNR performance as well as resolution capability of the estimator can be improved. For long-term estimation, we also develop a recursive evolutionary periodogram (REP) algorithm for updating the spectral estimate in an efficient way. Simulation results demonstrate that the proposed schemes significantly enhance the spectral resolution and tracking accuracy of the EP and DAE based methods, especially with low SNR and closely spaced moving sources.