Towards smart system architectures: A fuzzy MCDM-based evaluation of application mapping strategies

Application mapping strategies in Network-on-Chip (NoC)-based Multiprocessor System-on-Chip (MPSoC) are critical for achieving efficient communication and reduced energy consumption. Therefore, choosing the optimal mapping strategy is of significant importance. However, due to the numerous evaluation criteria, trade-offs, conflict, and criteria importance, the assessment and selection of mapping strategies remain a complex challenge. Despite the importance of this issue, current literature reveals a significant research gap in comprehensive comparative evaluations of these strategies using systematic and quantitative methods. Previous researchers recommended multi-criteria decision-making (MCDM) to address the issue of identity best mapping strategy. Remarkably, the literature has reported a paucity of evaluations of the optimal mapping strategies. The present study aims to determine the most effective application mapping strategies in certain situations by using fuzzy MCDM methods. The design and methods of this study involve two phases. The first phase involves the evaluation decision matrix, which is derived through the intersection of the evaluation criteria and the mapping strategies list. The second phase includes the proposed MCDM methods, namely the Weight Fuzzy Judgment Method with Triangular Fuzzy (Tr-WFJM) for determining the weights for the criteria of mapping strategies and Multi-Attributive Border Approximation Area Comparison (MABAC) to rank the mapping strategies based on the weight assigned. The findings of Tr-WFJM revealed that PIP Cost has the highest final weight (0.2326) and MPEG-4 Cost has the lowest weight (0.0887), respectively. In terms of the MABAC method, the Integer Linear Programming (ILP) is the most efficient mapping strategy. This study is exceptional because it provides academics and practitioners insight into reducing resources and energy consumption.