Optimization Algorithms for Design Space Exploration of Embedded Systems - Multi-objective Design Space Exploration of Multiprocessor SoC Architectures - page 51

Chapter 3
Optimization Algorithms for Design Space
Exploration of Embedded Systems

Enrico Rigoni, Carlos Kavka, Alessandro Turco, Gianluca Palermo,
Cristina Silvano, Vittorio Zaccaria, and Giovanni Mariani

Abstract This chapter is dedicated to the optimization algorithms developed in the
MULTICUBE project and to their surrounding environment. Two software design
space exploration (DSE) tools host the algorithms: Multicube Explorer and mod-
eFRONTIER. The description of the proposed algorithms is the central part of the
chapter. The focus will be on newly developed algorithms and on ad-hoc extensions
of existing techniques to face with discrete and categorical design space parameters
that are very common when dealing with embedded systems design. This chapter
will also provide some fundamental guidelines to build a strategy for testing the per-
formance and accuracy of such algorithms. The aim is mainly to build confidence in
optimization techniques, rather than to simply compare one algorithm versus another
one. The “no-free-lunch theorem for optimization” has to be taken into consideration
and therefore the analysis will look forward to robustness and industrial reliability
of the results.

3.1 Introduction

The optimization problems emerging in the design of embedded systems—and in
particular those addressed within the MULTICUBE project— are multi-objective and
characterized by the fact that all configuration parameters are discrete and possibly
categorical.
Having more than one objective to be optimized (maximized or minimized) im-
plies that the outcome of the optimization process is not a single solution but a set
of solutions. This set of solutions, which is called the Pareto front, represents the
trade-offs between the different (and possibly conflicting) objectives. A design is a
Pareto design (or point ) if it is not possible to improve one of its objective values
without deteriorating at least another one, as in Fig. 3.1 .

A. Turco ( )
ESTECO, Trieste, Italy
e-mail: [e-mail]