Document Type

Theses, Ph.D

Rights

Available under a Creative Commons Attribution Non-Commercial Share Alike 4.0 International Licence

Publication Details

Theses sucessfully submitted to the Technological University Dublin for the award of Doctor of Philosophy in February, 2009.

Abstract

In recent years the efficient and structured analysis of unknown CMOS integrated circuits (ICs) has attracted a lot of interest. Over the last decade different and non-invasive strategies have been developed to analyse unknown ICS. However, invasive procedures must always lead to the destruction of the system under investigation. Non-invasive approaches published so far have the disadvantage that ICs are analysed using very complex and very time consuming algorithms. The focus of this thesis is to develop a non-invasive and efficient procedure to determine fully unclassified digital CMOS ICs solely by their input-output behaviour. In this research, automata theory was used to develop algorithms to analyse and determine the behaviour of unknown ICs. A novel method to non-invasively identify the pins of the circuit using electrostatic discharge is presented. Then a division of the overall IC into different independent circuits is carried out using binary maximum sequences. Furthermore, a separation procedure was developed to quickly identify different behaviours thus, allowing to significantly minimise the following analysis. The main part of this research concentrates on the analysis of nonlinear unknown ICs which are most commonly used in ICs. Here, a novel procedure was developed capable of successfully determining the behaviour of the IC under investigation. The procedure introduced is able to automatically determine the number of states which did help to overcome some of the limitations of the traditional approaches. To demonstrate the correct operation of the algorithms developed a hardware analysis environment was also developed. The overall system presented in this research was implemented and tested using the IEEE ISCAS benchmarks. For every circuit analysed the behaviour was successfully determined. However, the methods developed in this research are not limited to the investigation of unknown CMOS ICs. They can also be used to non-destructively analyse structures, functions or behaviours in a wide variety of areas such as lexical analysis, pattern matching, communication protocols or software analysis.

DOI

https://doi.org/10.21427/D73606


Share

COinS