A Novel Design in Formal Verification Corresponding to Mixed Signals by Differential Learning

Abstract

The mixed signals exhibit a characteristics called loops composed of multiple feedbacks and thus it is not feasible for apply traditional testing methods for conduction sophisticated formal verification with higher accuracy accompanied by speedy response. We surveyed the current research work in this regards to find that there are open-end issues pertaining to mixed signal formal verification. This paper has displayed a novel formal verification procedure of the mixed signal utilizing differentially-placed neural network. An analytical demonstrating is given (i) an algorithm for generating multiple mixed signal comparing to feasible operational states of mixed signal circuits, (ii) algorithm for formal verification and (iii) algorithm for training. The accomplished result of comparative analysis demonstrates 98.7% of accuracy with speed response as compared to the current learning algorithms.