Goals

 

The proposed project has several goals. These goals serve as guidelines in the planification and scheduling of the work as well as providing a final objective at the conclusion of the development.

 

The recovery of the original signal from a set of measurements taken by using compressive sensing is an NP-Hard optimization problem, where one must recover an N-dimentional signal from M measurements with M much smaller than N. Many recovery algorithms have been proposed, being the most used the greedy algorithms. These algorithms work by iteratively choosing the local optimum value with respect to some heuristic.

 

Given the structure of the algorithms, and the inherent matrix operations which are needed to implement them, there are processing optimizations that can only be accomplished by exploring dedicated hardware structures. To take advantage of this, one can use field-programmable field arrays. 

 

As such, the main goals are:

 

• Study of compressive sensing recovery algorithms state-of-the-art

 

• FPGA implementation of Iterative Hard Thresholding for image recovery 

 

• Analysis of speed, recovery quality and power efficiency