合成孔徑雷達 (Synthetic Aperture Radar， SAR) 具有全天候與全天時成像能力，因此廣泛應用於地球觀測、軍事監測以及災害監控等領域。近年來，隨著高解析度成像需求提升，多通道SAR與長駐聚光燈模式 (Long Dwell Spotlight Mode) 等進階SAR技術逐漸受到重視。然而，這些進階模式會導致都卜勒頻寬增加，使系統容易因取樣率不足而產生方位向模糊與頻譜混疊問題。本文針對Displaced Phase Center Antenna (DPCA) 與Two-Step Processing Approach (TSPA) 兩種前處理演算法進行效能分析與硬體資料排程設計。於演算法部分，DPCA透過頻譜重建方法提升等效取樣率以抑制雙通道SAR系統之方位方向模糊與非均勻取樣之問題；TSPA則透過方位方向卷積提升等效取樣率，以消除長駐聚光燈模式下之頻譜混疊現象。而基於以上兩種演算法之特性，本文設計一種整合演算法使雷達可操作在雙通道常駐聚光燈模式下。此外本文亦設計了一種資料排程方式，使DPCA與TSPA可和RDA硬體進行介面的共用。首先，提出適用於雙通道DPCA系統之交錯式HBM資料排列方法，使系統可於單次Burst存取中同時取得雙通道資料，藉此使BRAM使用量由92.27%降低至71.06%，此外，亦提出HBM資料再利用之方法，藉此降低記憶體使用量與HBM存取需求。研究結果顯示，所提出之資料排程策略可有效提升記憶體使用效率、降低硬體資源需求，並使DPCA與TSPA得以共用RDA介面硬體架構，驗證其應用於SAR前處理硬體實現之可行性。;Synthetic Aperture Radar (SAR) provides all-weather and day-and-night imaging capability， and is therefore widely applied in Earth observation， military surveillance， and disaster monitoring. In recent years， with the increasing demand for high-resolution imaging， advanced SAR techniques such as multi-channel SAR and long dwell spotlight mode have attracted significant attention. However， these advanced modes enlarge the Doppler bandwidth， making the system more susceptible to azimuth ambiguity and spectrum aliasing problems due to insufficient sampling rates. This thesis focuses on the performance analysis， evaluation， improvement， and hardware data scheduling design of two preprocessing algorithms: Displaced Phase Center Antenna (DPCA) and Two-Step Processing Approach (TSPA). In the algorithmic aspect， DPCA improves the effective sampling rate through spectrum reconstruction to suppress azimuth ambiguity and nonuniform sampling problems in dual-channel SAR systems. TSPA， on the other hand， increases the effective sampling rate through azimuth convolution to eliminate spectrum aliasing in long dwell spotlight mode. Based on these two algorithms， an integrated hardware architecture for DPCA and TSPA preprocessing is proposed in this thesis. First， an interleaved HBM data arrangement scheme for dual-channel DPCA systems is proposed， allowing simultaneous acquisition of dual-channel data within a single burst access. As a result， the BRAM utilization is reduced from 92.27% to 71.06%. In addition， an HBM data reuse method is proposed to further reduce memory usage and HBM access requirements. The results demonstrate improved memory efficiency and reduced hardware resource requirements. The proposed strategy also enables DPCA and TSPA to share the same RDA interface architecture， showing its feasibility for SAR preprocessing hardware implementation.
