Mapping and Placement in NoC-based Reconfigurable Systems Using an Adaptive Tabu Search Algorithm

Abstract

Mapping and Placement still are big challenges in Networks-on-Chip (NoCs) design, due to the scalability, although several heuristics have been proposed to solve them. These problems belong to the class of Quadratic Assignment Problems (QAP). For NoC-based dynamically reconfigurable systems (NoC-DRSs), both mapping and placement problems present an additional complexity level due the reconfigurable layers/scenarios, being treated only by Genetic Algorithm meta-heuristics; however, several researches have described Tabu Search meta-heuristics as the best QAP solvers. This paper presents a formalization for the mapping and placement on 2D-Mesh FPGA NoC-DRSs, and provides as solver, a novel approach of adaptive Tabu Search, named Nav-adaTS. Results with a series of benchmarks are presented and compared to a basic adaptive Tabu Search and to the genetic algorithm implementation.