An optimized resource scheduling algorithm based on GA and ACO algorithm in fog computing

With the rise of Internet of Things (IoT) technology, fog computing has emerged as a promising solution for low-latency and real-time applications. As a highly virtualized platform, fog computing provides computing and storage services at the network edge to meet users' needs for latency-sensitive applications. However, resource scheduling is crucial in meeting customer demands and improving service quality. If the resource scheduling problem for large-scale service requests cannot be effectively solved, it will reduce resource utilization and decrease user satisfaction. Therefore, we propose a resource scheduling model called Normalization Processing to find the optimal pheromone for achieving the lowest total cost. The optimal resource scheduling result can be achieved by changing the ant pheromone concentration in the simulated foraging process. We also propose a resource scheduling algorithm called New Genetic Ant Colony Optimization (NGACO) Algorithm that a combination of the improved genetic algorithm (GA) and the improved ant colony optimization (ACO) algorithm. The GA is improved by incorporating a randomized initialization strategy, while the ACO algorithm is enhanced with the use of niche technology. NGACO algorithm introduces a pheromone update method optimization of three operators and a pheromone correction factor in the pheromone update rule. It can update pheromone generation by roulette algorithm. The NGACO algorithm effectively improves the exploratory power of the algorithm while ensuring initial population diversity. Additionally, we introduce a penalty mechanism to handle constraints, while the niche technology addresses the optimization problem of multimodal functions. The experimental results show that the NGACO algorithm demonstrates excellent resource scheduling performance, with a 14.7%, 25%, and 12.8% reduction in makespan, economic cost, and total cost, respectively, compared to the ACO algorithm. Furthermore, the load balancing is 34.7% higher than the ACO algorithm.