<p>The IoT is witnessing a quick expansion, and it has a massive strain on fog computing infrastructures. The issues of dynamic resource management and network topology control lead to high consumptions of energy, over consumptions of latencies, and high migrations of virtual machines (VMs). These inefficiencies are a major underminer of the efficiency of real-time applications, such as autonomous vehicles and smart grids. This paper presents a new hybrid concept and combines a Gabriel Graph-based energy-saving topology and a Firefly Optimization (FFO) algorithm to allocate the resource dynamically. The proposed solution first is a set of a stable and a low-energy honesty communication backbone which limits the computational search space of the FFO and helps to make the excellent VM placement and migration decisions. The developed simulations in Cooja network simulator with industry datasets prove to be superior over the baseline algorithm like FFD and ACO. Quantitative data involve reduction of energy consumption by 44.39 percent, VM migrations by 72.34 percent, and active hosts by 34.36. The FFO uses adaptive brightness with dynamic VM allocation on GG-pruned nodes and Levy flights with migration decisions which provides an <i>O</i>(<i>T</i> <i>n</i>) convergence rate even with dynamic workloads. These enhancements are notable with the help of improving energy efficiency, up to 40 percent reduction in operational costs when implementing in large scale, minimized VM migrations to enhance system stability, and reduction in the number of active hosts and make them much less valuable in the need to maintain hardware, hence making the framework of the IoT ecosystem thereof sustainable.</p>