Resource-Constrained Optimizations For Synthetic Aperture Radar On-Board Image Processing

Abstract

Synthetic Aperture Radar (SAR) can be used to create realistic and high-resolution 2D or 3D reconstructions of landscapes. The data capture is typically deployed using radar instruments in specially equipped, low flying planes, resulting in a large amount of raw data, which needs to be processed for image reconstruction. However, due to limited on-board processing capacities on the plane (power, size, weight, cooling, communication bandwidth to ground stations, etc.) and the need to capture many images during a single flight, the raw data must be processed on-board and then sent to the ground station efficiently as image products. In this paper we describe the processing architecture of the digital beamforming SAR (DBFSAR) of the German Areaospace Center (DLR) and the special steps that had to be taken to enable the on-board processing. We explain the required software optimizations and under which conditions their integration in the SAR imaging process leads to (near) real-time capability. We further describe the lessons learned in our work and discuss how they can be applied to other processing scenarios with limited resource availability.

BEAST Lab: A Practical Course on Experimental Evaluation of Diverse Modern HPC Architectures and Accelerators

“…In such cases, manual implementation of SIMD is often required, potentially necessitating restructuring the algorithms to tap into parallel processing capabilities. In this project, students extend the implementation of a sinc interpolation function [18] with OpenMP as well as manual SIMD and evaluate their solutions on several CPU platforms in BEAST. (3) In the last project, students work on a time series mining application, specifically matrix profile computation [17].…”

“…The projects cover the same type of topics and experiments but need more effort in programming, enabling multi-threading and various optimizations such as low-level manual vectorization. These projects target a selection of conventional and fairly new applications such as multigrid solver [20], matrix profile computation [17], and interpolation kernels [18].…”

Real-Time Capability of Dlr's Beamforming Synthetic Aperture Radar Processing Architecture

“…SIMD and OpenMP are techniques applied to optimize the run-time performance of code by exploiting the par-allel processing capabilities of modern CPUs. A detailed explanation of our optimizations is provided in [7], [8], and [9].…”

“…When the data is stored sequentially, Intel's SIMD instruction sets can be applied efficiently and thus create fast single-core kernels (components). Therefore, we used it, for example, to vectorize the interpolation required repeatedly in the processing pipeline [9].…”

Towards On-Board SAR Processing with FPGA Accelerators and a PCIe Interface

“…As a point of comparison, Ref. [13] presents image dimensions similar to the MMTI example; the article tackles on-board SAR processing using SIMD instructions with an Intel ® Core™ i7-3610QE (by Intel Corporation in Santa Clara, CA, USA) processor. The image dimensions are 7.5 km by 2.5 km with a resolution of 0.5 m. Note the large difference between the data weight in [11] (172 MB) and in the MMTI example (210 GB), as well as the image size, 120 m × 100 m in [12] in contrast to 7.5 km × 2.5 km.…”