Hybrid evolutionary optimization for efficient placement of IoT applications in fog computing environments

The dynamic and resource-constrained nature of fog computing has led to an increasing focus on enhancing application placement strategies. Recent developments in this field aim to address the limitations of traditional approaches and optimize the deployment of Internet of Things (IoT) applications in complex and heterogeneous fog environments. In this work, we propose a novel Collaborative Evolutionary Application Placement (CEAP) algorithm that integrates Differential Evolution (DE) with an elitist genetic-based scheme in a hybrid framework and employs a Pareto-optimal fitness function to preserve the multi-objective characteristics of the problem. The algorithm effectively explores different regions of the search space and generates a diverse set of high-quality solutions, reflecting multiple trade-offs between total cost and makespan. Experimental results demonstrate the robustness of the CEAP algorithm in terms of solution quality, diversity, and trade-off balancing across varying system workloads and system configurations. Compared to the Non-dominated Sorting Genetic Algorithm II (NSGA-II) and its parameter variants, the proposed CEAP algorithm achieves lower cost, makespan and deadline violations while producing a well-distributed Pareto-optimal solution set. Additionally, compared to the Local Search Drafting Niche Pareto Genetic Algorithm (LD-NPGA), an improved Niche Pareto Genetic Algorithm (NPGA) variant tailored for fog task placement, the proposed algorithm achieved up to 35% reduction in cost and 15% in makespan. It also outperformed weighted objective methods, including Time Cost-aware Scheduling (TCaS), Fog Service Placement Genetic Algorithm (FSPGA), Fog Service Placement Particle Swarm Optimization (FSPPSO), and the hybrid Genetic Algorithm–Flamingo Search Algorithm (GA-FSA), achieving 17% to 49% cost reduction and 10% to 25% makespan improvement across varying workload sizes.