A new approach for wavelength assignment in optical burst switching networks

In this paper, we propose a new approach for wavelength assignment in optical burst switching (OBS) networks. Unfortunately, existing priority based approaches do not efficiently use available information at the routers. Our approach employs learning-based wavelength assignments (LWA) in OBS networks. In order to strengthen LWA approach, we further propose two algorithms; LWA with preemption (LWA-WP) and dynamic burst aggregation (DBA) algorithms. While the former aims at reduced burst drops for wavelength conversion incapable routers, the latter aims at speeding up the learning phase of LWA algorithm. We show that the proposed learning based approaches profoundly decrease the burst drop ratio and increase the system performance.