Abstract:
Recent advancements in semiconductor technology have sparked a struggle among
researchers and manufacturers to best utilize the modern technology trends in
designing state of the art digital systems. Devices with small form factors, offering high
throughput and low power consumption are very much in demand. These factors have
actuated active research in the field of area efficient, low power high speed digital
system design. This research is an effort to contribute in this active research area by
adding a new dimension to digital design methodology. In addition to this, the research
also makes use of established digital design methodologies augmented with the
research studies outcome to produce novel designs around few exemplary applications.
The prime focus of this study is to explore Trace Scheduling Methodology and extracts
novel algorithm-to-hardware mapping features for efficient hardware design. Trace
scheduling is a topic under compiler design theory and is efficiently used to design
compliers for VLIW machines. The research, inspired by trace scheduling, introduces
the concept of efficient hardware design through identification of traces in the algorithm
and their mapping for optimal hardware affinity. The research work first investigates this
concept on a relatively simpler design such as an FIR filter in order to establish a link
between the two technology domains which are compiler theory and digital system
design. Later, the devised methodology is applied on applications from machine vision
and cryptography to design area efficient, low power, moderate data rate architectures.
The research presents novel hardware mapping of Peak Sorter and AdvancedEncryption Standard (AES) algorithm for moderately high data rate applications. The
designs offer a best area performance tradeoff. The utility of the technique developed in
this research can be found in mapping complex algorithms in Very Large Scale
Integrated (VLSI) circuits and digital design compilers