Analysis of blocking probability in WDM all-optical networks with non-Poisson traffic statistics

We study blocking probability in wavelength division multiplexing (WDM) optical networks. We consider two type of networks: with and without wavelength conversion, and two types of wavelength assignment policies: random wavelength assignment and first fit. We have shown that difference between values of blocking probability is significant for the cases when input flow characteristics are distributed according to memoryless and general distribution. We have found that values of blocking probability for models with general distribution are upperbounded by the values of memoryless distribution case. We have also considered a model with non-uniformly distributed destination node for an incoming connection request.