Intelligent Resource Allocation for Delayed Protection Based on Machine Learning in Elastic Optical Networks
Elastic Optical Networks; Routing, Spectrum Allocation, Modulation Tuning, Machine Learning
Currently, several Data Centers are interconnected worldwide, providing support to cloud services for processing and storing data generated on the internet. The Elastic Optical Network (EON) presents itself as an infrastructure capable of transmitting the high data rates generated in such Data Center networks, creating the Inter-datacenter Elastic Optical Networks (IDC-EONs). This is because the EON network has high bandwidth. In addition, it can operate at different transmission rates, increasing or decreasing the bandwidth according to demand, which makes the network ideal for supplying the world's data flow. In IDC-EONs, minor interruptions in transmission can result in the loss of a large volume of data. Some data can be critical, so their loss can cause harm to network users, hence the need to use protection strategies in IDC-EONs. From the study and analysis of data, it is possible to detect patterns and behaviors of the IDC-EON network, through the Machine Learning (ML) process, from the learning of a past experience. This work presents 3 resource allocation mechanisms for deferred protection in IDC-EON networks from this context. The mechanisms used in this work were based on data analysis and AM techniques to help decisions related to bandwidth allocation in optical paths in IDC-EON networks.