Intelligent Resource Scheduling for Edge-Integrated IoT Using Deep Learning

With the proliferation of connected devices brought about by the IoT, new obstacles have developed in the way of effective management of resources and processing of data. Autonomous cars, smart cities, healthcare, and industrial automation all rely on real-time processing and minimal latency responses. Some of the disadvantages of the conventional architecture in cloud computing are issues with the network congestion, latency, and scalability. But it is quite possible in edge computing that utilizes network bandwidth efficiently, has less latency, and includes computation in data origination. Rendering resource management at the edge challenging is due to constantly evolving IoT environments and the variability of workload and applications/devices’ demands. Drawing from deep learning approaches, this paper proposes an intelligent resource scheduling framework for IoT systems with edge integration. Thereby acting in real-time based on the existing condition of the network and the availability and needs of the connected devices, the framework adjusts the offloading and allocation of tasks. This means that when future workloads are predicted and resources properly assigned, the deep learning model adjusts task schedules. In conditions of scarce resources and a high rate of change in the environment, this method ensures the rational distribution of resources, reduces the load on the network, and significantly reduces the time required to complete work. Substantial experimental and simulation data show that the projected framework is better than the existing scheduling algorithms in terms of system outcome, power consumption, and time to task execution.