A Multi-Criteria Decision-Support Framework for Evaluating Fog Computing Challenges and Service Models

Abstract

Fog computing extends cloud computing by bringing processing, storage, and networking resources closer to end devices and data sources. This architecture can reduce latency, improve response times, decrease bandwidth consumption, and support applications that require rapid decision-making and the processing of large volumes of data. However, its adoption is affected by technological, operational, economic, and security-related challenges. This study develops a multi-criteria decision-making model to evaluate the principal challenges of fog computing and identify the most suitable fog computing service model. Seven criteria are considered: latency requirements, bandwidth efficiency, security and privacy, scalability and flexibility, energy efficiency, cost effectiveness, and network reliability. Six service models are evaluated as alternatives: Infrastructure-as-a-Service, Platform-as-a-Service, Software-as-a-Service, Data-as-a-Service, Communication-as-a-Service, and Edge-as-a-Service. The proposed framework integrates the Entropy and CoCoSo methods within a crisp decision-making environment. The Entropy method is used to determine objective criteria weights, while the CoCoSo method is employed to prioritize the alternatives. The results indicate that network reliability is the most influential criterion, with a weight of 0.2636, followed by energy efficiency with a weight of 0.2343. Security and privacy receives the lowest weight of 0.0360. Among the evaluated alternatives, Software-as-a-Service achieves the highest ranking, followed by Data-as-a-Service and Infrastructure-as-a-Service. Comparative analysis using the WASPAS and TOPSIS methods and sensitivity analysis under five scenarios are conducted to examine the robustness and stability of the findings. The results provide practical guidance for researchers, service providers, and policymakers involved in fog computing adoption and implementation across distributed, data-intensive, and time-sensitive computing environments in practice.