Although filterless optical networks are expected to require a larger wavelength usage than WSS-based networks because wavelengths follow trees instead of point-to-point paths, filterless networks can reduce capital expenditure by avoiding the usage of expensive WSSs. In addition, filterless networks may provide more robustness to faults thanks to the exploitation of the broadcast-and-select technique, they support multicast capabilities, and may also speed up the set up delay of lightpaths (being the set up time often dominated by WSS configuration).
The authors clearly detail the advantages, the issues and the constraints (e.g., wavelength tree) rising from the concept of filterless networks. Then, assuming the filterless-network constraints, the authors address the GMPLS control plane implementation, as well as the Path Computation Element (PCE) architecture. An impairment-aware routing and wavelength assignment (IA-RWA) performed by two PCEs, one devoted to RWA (based on a load balancing strategy) and the other one to the impairment validation, is proposed. Link interconnection discovery and advertising issues are addressed as well as the management of unfiltered channels. Finally, a performance evaluation analyzes the proposed filterless network implementation by studying performance parameters such as the wavelength usage in several network topologies with different nodal degrees.
The authors also provide valuable suggestions on future works, such as the study of optical networks with sparse usage of optical filters, looking at a trade-off between costs, set up time and wavelength usage.
You must log in to add comments.