|
Abstract : |
We propose a new class of decision diagrams called canonical cube transformation binary decision diagrams (canonical TBDD's), which is an extension of TBDD's proposed by Meinel et al [11, 3]. The core idea of TBDD's is to transform a function to another function in a new domain by an injective mapping and to represent the transformed function in a standard OBDD. If the new domain is larger than the original domain, canonicity is lost, which makes combinational verification difficult. In this paper we show that canonicity can be maintained by characterizing the care set of the new domain. Transformations of practical interest which guarantee polynomial size canonical TBDD's are introduced. We also give a new interpretation of TBDD's, which leads to an effective heuristic for extracting promising transformations automatically from highlevel specifications. Finally a combinational verification technique using canonical TBDD's is proposed, the effectiveness of which is justified by verifying the hidden weight bit function. 1, |
