Abnormality Detection Using Power Rising and Descending Signature (PRIDES)

Abstract

Listen

Translate article

This paper presents a new paradigm for abnormality detection using a novel power signature that characterizes the rising and descending patterns of energy consumption. The proposed methodology includes a low-overhead power signature generation circuit, computation-light analysis methods, and optimal generation of the golden signature used in the analysis. The proposed power signature generation circuit is designed using 90 nm CMOS technology, and its operation is validated via circuit simulations. The effectiveness of the proposed method in detecting the insertion of potentially malicious code is demonstrated with data obtained from hardware experiments and circuit simulations.