Task 3 - Hardware acceleration for imaging algorithms

POLITO , supervisor Prof. M. Casu, ESR M.A. Mansoori.

The task will identify a comprehensive set of computational kernels that are recurrent in microwave imaging algorithms (in particular those considered and/or developed in WP3). First, these kernels will be implemented in a high-level programming language to create a benchmark implementation. Then, the best hardware platform will be selected to accelerate the execution (compared to the reference) of these kernels under various constraints. Finally, the computational kernels will be implemented on the selected hardware platform.

Objectives:

• Identify a comprehensive set of computational kernels that are recurrent in the microwave imaging algorithms employed in the biomedical field.
• Create and benchmark a reference implementation of these kernels in a high-level programming language
• Evaluate the best hardware platform (e.g. multicore CPU, embedded CPU, GPU, Field-Programmable Gate Array (FPGA), dedicated Application-Specific Integrated Circuit (ASIC)) for the accelerated execution of these kernels under various constraints (e.g. cost, power, and form factor)

Expected Results:

• An optimized library with the most used computational kernels for microwave imaging
• An optimized hardware implementation of these kernels targeting a specific hardware platform