Enhanced Lemurs Optimizer for Multi-Objective Task Scheduling in Heterogeneous Fog–Cloud Architectures

Task scheduling plays a central role in Fog–Cloudcomputing systems, where heterogeneous resources and latencysensitiveworkloads require intelligent distribution of tasks toensure efficient operation. However, achieving optimal schedulingis challenging because the problem is NP-hard and involvescompeting objectives—minimizing makespan, reducing energyconsumption, and lowering processing cost. Existing heuristic andmetaheuristic methods often struggle to maintain a proper balancebetween exploration and exploitation, resulting in prematureconvergence or suboptimal solutions. This paper proposes anEnhanced Lemurs Optimizer (ELO) that integrates a problemawareheuristic initialization with the adaptive search dynamics ofthe original Lemurs Optimizer (LO). The heuristic phase generateshigh-quality starting schedules by considering execution time,power efficiency, and cost trade-offs, while the LO refinementphase stabilizes convergence through controlled exploration andexploitation. Extensive simulations across workloads of 100–400tasks demonstrate that ELO consistently outperforms benchmarkalgorithms. Compared to GA, ELO achieves up to 54% lowermakespan, up to 52% lower energy consumption, and up to 11%lower processing cost, while also exhibiting substantially morestable convergence behavior. Wilcoxon signed-rank tests confirmthat these improvements are statistically significant across allexperimental settings. These results highlight the potential of theELO as an effective and scalable scheduling framework for realworld Fog–Cloud systems.