.png){kind=icon}
Return
Tiled Fine-Grained GPU Parallel Acceleration Method for SAR Echo Simulation Based on the Concentric Circle Algorithm
.png){kind=icon}
PRE
AI
DOI：10.1109/lgrs.2026.3727065.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
With the increasing resolution and scene scale of synthetic aperture radar (SAR) systems, echo simulation requires substantially improved computational efficiency. Existing GPU-based, concentric-circle methods (CCMs) suffer from performance degradation caused by frequent atomic operations during concurrent accumulation into identical range cells. This letter proposes a tile-based, fine-grained GPU parallel SAR echo simulation method based on the concentric-circle echo model. By exploiting the separability of range-dependent modulation terms via the fast Fourier transform (FFT), echo generation is reformulated into a range-cell-prior 2-D tile structure over the azimuth–range domain. Coherent accumulation is performed locally within each tile, while the coherent processing interval (CPI) is further partitioned into independent azimuth–time blocks to reduce synchronization overhead and improve memory locality. Experimental results show that a $2048\times 2048$ SAR echo matrix can be generated in 0.532 s, achieving significant acceleration while preserving imaging accuracy and physical consistency.
Keywords:
Echo simulation
fine-grained parallelism
GPU acceleration
real-time imaging
synthetic aperture radar (SAR)
Journal
I
IF:
4.4
Papers:
714
Citations:
0
Organization
Cited Papers
No cited papers available