Revolutionizing optical burst switching networks with dual auto net and marine swarm optimization techniques

Optical Burst Switching (OBS) offers a promising solution for efficient bandwidth utilization in optical networks. This study aims to enhance burst assembly and scheduling in OBS networks using deep learning and optimization techniques. The research begins with data collection, focusing on key OBS network parameters such as packet counts, burst sizes, and traffic patterns. The DualAutoNet model, incorporating autoencoders, Convolutional Neural Networks (CNNs), and Recurrent Neural Networks (RNNs), is then employed to optimize burst assembly. For optimal channel scheduling, a novel hybrid optimization method, the Marine Swarm Optimization Algorithm (MSOA) which combines Tuna Swarm Optimization (TSO) and Tunicate Swarm Algorithm (TSA) is introduced. Additionally, a multi-objective optimization-based route queuing protocol is developed, accounting for latency, energy consumption, throughput, and distance. The MSOA is utilized to determine the best routes for efficient network resource management in OBS networks. The proposed model's performance is compared to existing methods, including Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Tunicate, and Tuna algorithms. Implemented using MATLAB, key performance indicators such as energy consumption, network lifetime, throughput, and packet delivery ratio are evaluated under varying node conditions. This paper presents a detailed comparative analysis of the results, demonstrating the proposed model's superiority in reducing latency, increasing throughput, and minimizing packet loss.