AI and ML-Driven Optimization of Telecom Routers for Secure and Scalable Broadband Networks

Abstract

As the core access network technology of tomorrow's optical Internet, optical roaming- and wavelength-division multiplexed passive optical networks are considered an effective candidate to meet the continuous growing capacity requirements. With the intention of supporting the future bandwidth-intensive applications, an improved estimated capacity model and an engineering-optimization-based resource-allocation scheme are proposed and developed for secured and cost-effective optical roaming- and wavelength-division multiplexed passive optical networks. The capacity model takes the mutual interference of the concurrent upstream transmissions and the optical-filtered attenuation into account for measuring the viewable capacity of ONU, which is the theoretical maximum data-rate of an ONU. On the basis of the minimal and unequal arrangement of the WDM/OFDM resources, a pluggable resource-allocation scheme that includes filter insertion, signal auditing, and resource allocation is proposed to ensure the member ONUs sufficient bandwidth to accommodate their requested transmission-rate. To avoid the potential link-layer vulnerabilities, an imitated data traceattack-agnostic OAM-encapsulation networking scheme is proposed to secure the tunnelling, forwarding, and retrieval of the active reflective multi-cast signals in optical roaming- and wavelength-division multiplexed passive optical networks. The scheme mitigates the potential network-wide outages caused by the inadvertent propagation of the counterfeit demand-bursts.