Engineering of an Effective Automatic Dynamic Assertion Mining Platform

Abstract

Several approaches exist for specification mining ofhardware designs, both at the RTL and system levels (e.g, TLM).These approaches mine assertions that specify the behavior ofthe design. Some of the techniques require the source codeitself while others can extract assertions directly from simulationtraces. The performance of some approaches is highly dependenton the number of simulation traces/use cases while there existapproaches which can extract assertions from a limited numberof simulation traces. Apart from this aspect, the core of eachassertion miner is different from the other ones. Some useexpression templates to define assertions while some are basedon the static analysis or information flow analysis. Unfortunately,it has been rarely considered which of the current approachesare more effective in describing functionality of particular typesof designs. Thus, in this work, we analyze assertion minerswhich are template based and dynamic dependency graph based,respectively. We generate assertions from both approaches. Theevaluation considers fault analysis on both assertion sets ofextracted assertions. Moreover, both sets are combined and faultanalysis has been applied on them. Experimental results showthat each set approximately detects the same number of faultswhile when the two sets are combined the number of detectedfaults increases. Finally, a new, more efficient architecture for aneffective assertion miner has been developed based on the studyin this work.