Cross-layer path computation for dynamic traffic grooming

Abstract

Traffic grooming in wavelength division multiplexing (WDM) optical networks is an operation to consolidate multiple low-speed traffic streams onto high-speed lightpaths. Traffic grooming of dynamic traffic over a dynamic logical topology, where lightpaths (logical links) are dynamically set up and torn down, is called dynamic traffic grooming, where a major task is to find a path (a list of adjacent lightpaths) in the dynamic logical topology to carry an incoming call. Dynamic traffic grooming is a significant issue in WDM optical networks due to the customer's need of the bandwidth-on-demand service. This paper studies path computation in dynamic traffic grooming, using various link cost functions. We propose a path computation approach utilizing the resource usage information in both logical and physical layers. We have evaluated the performance of this cross layer path computation approach, compared with the approach using the resource usage information in the logical layer only. The numerical results show that the cross layer path computation approach can achieve 10% - 20% lower blocking probability.