A Data Representation and Resource Allocation Approach for Intelligent Network Architecture

In order to support the digital, intelligent, and intelligent operation and maintenance of the next-generation optical network and the smooth evolution and upgrade of the network architecture, this paper conducts research on the optical network virtualization architecture technology that supports cloud-network integration, builds a cross-domain layered virtualization architecture, and studies cloud network resources. Unified representation technology and resource allocation technology to support large-scale AI distributed training scenarios, built a cross-domain ”IP + optical” experimental environment and calculated optical performance model errors by monitoring changes in computing, storage, and transmission resources in distributed optical performance monitoring scenarios. It is less than 0.5dB, which verifies the feasibility of the virtualization architecture. Finally, the development trend of network intelligence is discussed at the end of this paper.