Algorithm-Architecture Matching for Signal and Image Processing

Mobile devices are moving into fields such as cars, healthcare, telecom, multimedia and entertainment, largely because signal and image processing keep improving alongside greater computing power. Building these applications usually proceeds in three separate stages: studying the algorithms in theory, studying the target hardware, and then implementing the two together. That sequential approach is struggling under tight schedules and demanding performance requirements. Algorithm-Architecture Matching responds by analysing algorithmic and architectural questions at the same time, weighing several design constraints and optimizing both sides in ways that would be out of reach if they were handled apart. The need for such methods grows with emerging work such as advanced mobile communication, graphics on sub-micron processes, and 3D integrated circuits.

The volume gathers selected contributions from industry and academia covering the approach from sensor design through to architecture. Its range of algorithms includes signal, communication, image, video and 3D graphics, and the target platforms run from FPGAs to multiprocessor systems. Several synthesis and resource management techniques for design optimization are described and applied to a number of algorithms. Its stated audience is system designers, developers of electronic design automation tools, and engineers interested in optimizing digital systems that run advanced algorithms.