Research on Resource Continuous Scheduling Method Based on Multi-Objective Optimization for Smart Grids IoT Networks

In the IoT enabled smart grids networks, the resource is highly restricted and it is impossible to complete complex tasks such as multi-dimensional data processing and AI model training. Under this circumstance, introducing microservices to the IoT network can enhance the collaboration of edge nodes with computing function. However, dynamic resource rescheduling in this environment remains a key challenge. This paper proposes a scheduling scheme that assesses resource utilization through real-time monitoring and introduces a live migration mechanism to ensure that service instances can continue to run uninterrupted during adjustments. Secondly, to optimize the allocation of various computing resources in the network and further enhance the overall network performance, this paper proposes an improved Non-dominated Sorting Genetic Algorithm II (NSGA-II). The improved algorithm is specifically designed for multi-objective optimization problems, not only capable of simultaneously addressing the often conflicting goals of balancing computing resource distribution and reducing service impact latency, but also enhancing the ability to find solutions that meet multiple constraint conditions. Simulation results confirm that the proposed algorithm significantly improves comprehensiveness in dealing with the complex multi-objective optimization problem of resource balance and cost control, and can efficiently search for the optimal solution set.