An integrated Fuzzy-BWM, Fuzzy-LBWA and V-Fuzzy-CoCoSo-LD model for gateway selection in fog-bolstered Internet of Things

Abstract

This study proposes a fog gateway selection strategy based on a hybrid Multi-criteria Decision-Making (MCDM) optimization model to avoid overwhelming the fog–cloud network in fog computing architecture. The study suggests that only authorized fog gateways should communicate with the cloud to prevent prolonged response times and significant data loss. To prioritize the selection criteria, the proposed model integrates subjective weights obtained from Fuzzy Best–Worst Method (Fuzzy-BWM) and Fuzzy Level Based Weight Assessment (Fuzzy-LBWA) using an extended nonlinear weighted synthesis approach. A Vector-normalized Fuzzy Combined Compromise Solution with Later Defuzzification (V-Fuzzy-CoCoSo-LD) model is used to rank the alternative gateways. The study constructs a taxonomy of selection criteria and finds that Quality-of-service and Level-of-security are the most important criteria with weights of 0.3731 and 0.3452, respectively. The hybridized weighting method addresses the weaknesses of individual methods. The study also conducts sensitivity analysis of the impact of fuzzy weighting parameters α̃ and β̃ on the decision-making process, and the results show that the most viable alternative remains the same regardless of the changes. The sensitivity analysis of the most sensitive criterion as well as comparative and correlation analyses also validate the effectiveness and robustness of the proposed model. The study concludes that the vector normalization method is the best for Fuzzy-CoCoSo MCDM model, with Spearman’s and Pearson’s correlation coefficient mean values of 0.9404 and 0.9868, respectively, justifying the choice in this study. Finally, the study provides design implications for the practical application of the proposed model in fog computing services. The proposed model can assist fog service providers in making informed decisions, and this study presents a case study to demonstrate the effectiveness of the model.