Ant behavior may help semiconductor industry to verify design
One of the hardest problems in validating designs is getting the system to reach all the possible states it will experience during use. The hard-to-reach “corner cases” are compared to finding a needle in a universe of haystacks – or finding a particular state within a space of more than 103000 states.
Michael Hsiao’s team has developed an approach that finds many of the hard-to-reach states where existing methods fail. Their heuristic is based on ant colony optimization, biologically inspired by swarm-intelligence.
In their model, artificial ants start from their nest (the initial state) and attempt to find a path to reach food (the target state). In so doing, all the ants will converge to the path most likely to reach the target state.
“This is an important problem to solve,” Hsiao said. “Finding these hard-to-reach spaces will make design validation and verification more robust and more efficient.” Up to 70 percent of the effort of the semiconductor industry is spent checking if the implemented circuit conforms to the design intent, he noted.