Optimizing Task Offloading Efficiency in Non-Orthogonal Multiple Access-Enabled Edge Computing Networks Using Deep Reinforcement Learning

Edge computing shows potential, in supporting applications that require computation in wireless networks. Alongside advancements in edge computing, nonorthogonal multiple access (NOMA) has emerged as a way to improve spectrum efficiency. This study discusses an edge computing system where various edge devices utilize NOMA to coordinate their tasks with a central edge node that has computational resources for task processing. This framework involves coordinating task scheduling allocating resources and managing power distribution to minimize costs, which include delay and energy consumption. Understanding the balance between energy usage and delay is essential, in this context. It is challenging to Find the best practice for this complex combinatorial optimization problem is a real challenge. To deal with this, we offer an online knowledge-based optimization system leveraging Deep Reinforcement Learning.Our simulation results investigate and compare with the previous works on the basis of reducing the cost, resource allocation, task scheduling and proper power distribution.