Modeling and Analysis of WDM OPS Employing Tunable Converter Sharing under Self-Similar Variable Length Packet Traffic

Wavelength division multiplexing (WDM) switching has been emerging as the dominant technology in optical networks. In this paper, we study the switching performance of WDM optical packet switch (OPS) equipped with tunable wavelength converter sharing dedicated to each output fiber line under self-similar variable packet length traffic. In this study, we model WDM OPS employing tunable wavelength converters as a queueing system. In our model, packets arriving at OPS are assumed to follow Markov modulated Poisson process (MMPP) that emulates self-similar traffic, and service times (packet lengths) are exponentially distributed to cope with variable packet length. We investigate packet blocking probabilities against traffic load for different wavelength conversion capabilities under self-similar traffic. We propose an efficient analytical method and the results are validated with simulations. We present the computational complexity of the proposed analytical method. Our analysis is more significant in dimensioning WDM OPS to guarantee quality of service (QoS).