OptFog: Optimized Mobility-Aware Task Offloading and Migration Model for Fog Networks

Due to the increased usage of mobile devices, there are now diverse application needs that favor the use of Fog computing architecture over the Cloud's centralized architecture. Fog Computing arises in this circumstance, which moves certain processing resources to the edge or among themselves and provides low-latency access to networked devices. However, minimizing the number of migrations to utilize the limited resources at the fog devices is essential and challenging. In this regard, this paper presents OptFog (optimized mobility-aware predicted task migration model for fog-assisted networks) as a migration assistance model. Three machine learning models are employed to anticipate mobility and make proactive migration decisions. The data used in OptFog is from active users provided Luxembourg SUMO Traffic (LuST) from the city of Luxembourg. Machine learning models like Decision Tree, k-Nearest Neighbors (k-NN), and Random Forest are evaluated to estimate the user's future position. Thereafter, the best approach is determined by comparing them. The Random Forest model produces a significant result, with an overall accuracy of 97.47%. OptFog shows the overall time, downtime and energy consumption when employed with a prediction-based migration strategy and also analyzes the most elevated scenarios for task migration. Simulations have shown that migration can result in satisfactory effectiveness for supporting mobility in fog-assisted networks.