A research paper titled SAR Image Processing using GPU has been selected for oral presentation at the 40th COSPAR (Committee on Space Research ) Scientific Assembly, Moscow, Russian  Federation on August 2- 10, 2014. The paper is authored by Dr. Shanmugha Sundaram G. A., Assistant Professor at Department of Electronics & Communication Engineering, Amrita School of Engineering and Amrita Center for Computational Engineering & Networking, Coimbatore. 

The detailed version of the paper has also been invited for publication in the COSPAR publication Advances in Space Research (ASR).

The talk is scheduled under the session, Space Studies of the Earth’s Surface, Meteorology and Climate. The objective of the session includes developing the global capability to understand planet Earth, predict changes and mitigate negative effects of global change on its population. The symposium highlights the essential importance of space for monitoring Earth as a whole and understanding the impact of human activity on our planet.

COSPAR, established by International Council for Science, aims to promote scientific research in Space. Every second year COSPAR organizes a general assembly with emphasis on the free exchange of results, information and opinions, and providing a forum, open to all scientists, for the discussion of problems that may affect space research.

“The paper helped in complementary and supplementary Earth monitoring activities using Graphics Processing Unit (GPU),”stated Dr. Shanmugha Sundaram.

“The paper discussed the relevance of High Performance Computing (HPC) for SAR data processing using an off-the-shelf GPU over CPU and quantitative estimates on SAR image processing performance comparisons using both CPU and GPU were also provided,”he further explained.

The Amrita Center for Computational Engineering and Networking (CEN) has its own GPU-accelerated computing facility based on NVIDIA.

Syam Sujith Maddikonda, II year, M. Tech. Remote Sensing and Wireless Sensor Networks is undergoing the project activity of the paper under the guidance of Dr. Shanmugha Sundaram G. A. This is the second time that Amrita paper was selected for COSPAR, the first one being in the 39th COSPAR SA 2012 presented by KAS Joseph, II year M. Tech. Remote Sensing & Wireless Sensor Networks of CEN presented a paper  at Mysore, India.

Given below are some excerpts from the paper, SAR Image Processing using GPU:

Synthetic Aperture Radar (SAR) has been extensively used for space-borne Earth observations in recent times. Unlike optical systems that rely on reflected solar radiation or thermal radiation emitted by Earth, imaging SAR instruments work independently of light and heat. SAR is an active system that transmits a beam of radiation in the microwave region of the electromagnetic spectrum.SAR can provide day-and-night imagery of Earth. In addition, clouds, fog and precipitation do not have any significant effect on microwaves, so images can also be acquired independent of weather conditions.

The imaging SAR data is highly computation-intensive in terms of processor and memory usage during phases of routine I/O, handling and archiving. A conventional CPU consists of one or few processor cores optimized for sequential serial processing while a graphics processing unit (GPU) consists of thousands of smaller, more efficient cores designed for handling multiple tasks simultaneously.

GPU-accelerated computing offers unprecedented application performance by offloading compute-intensive portions of data processing to the GPU, while the remainder of the code still runs on the CPU. From a user’s perspective, applications simply run significantly faster. This division of labor vastly simplifies the SAR data processing activity, and renders end-user-friendly information of scientific grade.

May 31, 2014
Amrita School of Engineering, Coimbatore