A novel blocking analysis method in wavelength-routed networks

In wavelength-routed networks (WRN), blocking probabilities (BP) of on-demand requests are an important performance measure. The computational complexities of analyzing BP in conventional ways grow exponentially with the number of wavelengths, W. Therefore, analytical approaches are impracticable even in very small DWDM networks which employ tens or hundreds of wavelengths per fiber. Though simulations are always used to evaluate performance, analytical approaches are preferable if they have good accuracy, feasible computational complexities and deep insights into the variance rules of network states. We propose a novel analytical method for BP, whose computational complexity is independent of W, so the analytical approach becomes practicable. Besides, for the first time, we applied this method to study the success probabilities of forward reservation (i.e., label suggestion in GMPLS) when global wavelength usage information is not available. It is shown that our approach is of great importance in analyzing the performance of forward reservation.