Transmission analysis by using federated machine learning model in optical access networks based multi-agent communication and routing system

Due to the Internet's and communication systems' rapid technological and service development, communication networks have been impacted by a rise in intricacy. Optical networks, which are vital components of both the core and access networks in communication networks, face major obstacles due to the complexity of the system and the requirement for manual effort. To overcome the current limitations and address the issues with future optical networks, it is essential to install greater intelligence capabilities to enable autonomous as well as flexible network operation. The present study delivers a novel approach for transmission analysis-routing-improved multiagent communication using optical access networks. An optical access network is used here, and routing is handled dynamically using multi-agent communication. A federated convolutional component neural network is then used to analyse the data flow. Throughput, latency from end to end, lifetime of the network, route loss, and energy consumption are all measured experimentally and analysed. When designing the routing protocol, we first model the network as a decentralised multi-agent system and factor in factors like residual energy and connection quality. As a result, the network is better able to adjust to changes and the lifespan of the network may be prolonged. The proposed solution increased network throughput by 97%, decreased end-to-end latency by 54%, prolonged the life of the network by 6%, reduced route loss by 59%, and reduced energy consumption by 55%.