Multi-Network Task Offloading with Improved Recording Neural Network

Abstract:
For the current development of edge computing, it has become a new trend for multiple wireless devices to form a network to connect to the entire system. At the same time, task offloading in this scenario is partially different from traditional edge computing. For multiple systems, how to efficiently allocate tasks has become the primary problem that needs to be solved in this scenario. This paper proposes an improved task offloading algorithm based on neural network. By using neural network to learn the process of task offloading, and at the same time calculate the cost after each decision, the decision that makes the cost decreasing will be recorded, the decision with the original task size is used as the input of the neural network together. Experiments show that the trained neural network can quickly and efficiently obtain the offloading result after each task input while ensuring the lowest delay and energy consumption.