Flexible and efficient network slicing scheme based on resource awareness and traffic prediction for Fi-Wi access network

Abstract

Optical access networks are widely recognized as a pivotal technology for addressing the last-mile connectivity challenges. However, the continuous emergence of novel applications within the industrial Internet introduces unprecedented challenges to these networks. This paper proposes a resource-aware dynamic slicing scheme for Fi-Wi access networks. This proposed scheme adheres to the ISA-95 standard and seamlessly integrates network slicing technology into optical access networks. The scheme uses a resource-aware model to select the optimal path and reduce processing latency. LSTM-Attention is introduced for traffic prediction, reducing the delay in slice adjustments caused by large-scale traffic. Simulation results show that this scheme can efficiently select the optimal path, reserve more resources for services, and has a lower bandwidth blocking rate compared to traditional DES and LSTM methods.