Edinburgh Napier University

  In this thesis, a variety of algorithms for synthesis and optimisation of combinational and sequential logic circuits are developed. These algorithms could be part of new commercial EGAD package for future VLSI digital designs. The results show that considerable saving in components can be achieved resulting in simpler designs that are smaller, cheaper, consume less power and easier to test.
The purpose of generating different sets of coefficients related to Reed Muller (RM) is that they contain different number of terms; therefore the minimum one can be selected to design the circuits with reduced gate count. To widen the search space and achieve better synthesis tools, representations of Mixed Polarity Reed Muller (MPRM), Mixed Polarity Dual Reed Muller (MPDRM), and Pseduo Kronecker Reed Muller (PKRO RM) expansions are investigated. Efficient and fast combinatorial techniques and algorithms are developed for the following:

- Bidirectional conversion between MPRM/ MPDRM form and Fixed Polarity Reed Muller forms (FPRM)/Fixed Polarity Dual Reed Muller forms (FPDRM) form respectively. The main advantages for these techniques are their simplicity and suitability for single and multi output Boolean functions.
- Computing the coefficients of any polarity related to PKRO_RM class starting from FPRM coefficients or Canonical Sum of Products (CSOP).
- Computing the coefficients of any polarity related to MPRM/or MPDRM directly from standard form of CSOP/Canonical Product of sums (CPOS) Boolean functions, respectively. The proposed algorithms are efficient in terms of CPU time and can be used for large functions.

For optimisation of combinational circuits, new techniques and algorithms based on algebraic techniques are developed which can be used to generate reduced RM expressions to design circuits in RM/DRM domain starting from FPRM/FPDRM, respectively. The outcome for these techniques is expansion in Reed Muller domain with minimal terms. The search space is 3`" Exclusive OR Sum of Product (ESOP)/or Exclusive NOR Product of Sums (ENPOS) expansions.
Genetic Algorithms (GAs) are also developed to optimise combinational circuits to find optimal MPRM/MPDRM among 3° different polarities without the need to do exhaustive search. These algorithms are developed for completely and incompletely specified Boolean functions. The experimental results show that GA can find optimum solutions in a short time compared with long time required running exhaustive search in all the benchmarks tested.
Multi Objective Genetic Algorithm (MOGA) is developed and implemented to determine the optimal state assignment which results in less area and power dissipation for completely and incompletely specified sequential circuits. The goal is to find the best assignments which reduce the component count and switching activity simultaneously. The experimental results show that saving in components and switchingactivity are achieved in most of the benchmarks tested compared with recentlypublished research. All algorithms are implemented in C++.