A novel two-layer wavelength conversion (TLWC) scheme, for reducing the implementation cost of an optical switching (OS) node, is presented in this paper. The study stems from practical observation that wavelength converters are expensive. The paper thus presents a new OS node architecture that uses significantly less wavelength converters without compromising on drop performance. In this TLWC system, partial wavelength converters (PWCs), which can only convert wavelengths within a limited range, form the first layer; thereafter, complete wavelength converters (CWCs), which can convert wavelengths within all input ranges, form the second layer. When new optical data needs a wavelength converter, PWC resources are first provided; if PWC resources are not available, CWC resources are used. For the case of the CWC layer, a sharing-mode scheme called the share per fiber (SPF) mode for improving the utilization of the limited number of wavelength converters is proposed. Subsequently, a two-dimensional Markov-chain model of TLWC-SPF is presented to evaluate its performance. Numerical-simulation results are presented to verify the accuracy of the analytical model. The analysis shows that the cost of the OS fabric of TLWC-SPF is similar to an OS node employing a limited number of CWCs (i.e., CWC-SPF). However, in terms of wavelength-converter savings, the TLWC-SPF architecture provides 50% (in heavy load) savings compared to full wavelength conversion (FWC) optical nodes, while 40% savings is achieved compared against CWC-SPF nodes.
© 2006 IEEE
Hailong Li and Ian Li-Jin Thng, "Cost-Saving Two-Layer Wavelength Conversion in Optical Switching Network," J. Lightwave Technol. 24, 705- (2006)