Flow Scheduling in Optical Flow Switching Networks Under Transient Conditions

We study flow scheduling to lightpaths in an optical flow switching network where flow arrivals are modeled as a time dependent Poisson process and their holding times are Pareto distributed. Scheduling decisions are taken at fixed time intervals (larger than 100 ms) in a central node and scheduled flows cannot be interrupted before their completion. The scheduling problem is represented as a discrete time Markov decision process with flow blocking probability objective function over a finite horizon. We derive three lower bounds to the objective function and propose several schedulers, with and without fairness requirements. The schedulers performance are evaluated under both, static and limited dynamic routing, by emulating the algorithms on random networks topologies for 7200 seconds. The main result is that our proposed max-min fair scheduler with limited dynamic routing significantly outperforms all schedulers with static routing. Furthermore, its blocking probabilit y is close to the lower bound for static routing.