A novel analytical model for dynamic waveband switching

We present a novel framework to analytically model dynamic waveband switching in a multi-granular optical network. The scalable solution consists in modeling each potential carrier of waveband tunnels independently by a Markov chain while modulating the rate of critical transitions, i.e. reserving a new waveband tunnel, by the waveband setup availability computed from the solution of other potential carriers. An iterative procedure is repeated to obtain a consistent numerical solution all over the network. To get an accurate solution we present a novel method to solve the problem of link load correlation in the analysis of circuit-switched networks. Analytical and simulation results are presented to show the effectiveness of the proposed model. The model is also applied to evaluate the performance of waveband cross-connects when introducing waveband conversion.     

Analytical calculation of blocking probabilities in WDM rings with the first-fit algorithm

The estimation of blocking probabilities in optical path networks has been investigated mainly for the case in which wavelengths are chosen randomly among the available ones on a route. Even though algorithms that unbalance the load among wavelengths usually have a superior performance when compared with the random algorithm, the modeling complexity of these better-performance algorithms often limits their analytical analysis. This article proposes a per-wavelength plane blocking probability estimation for the first-fit algorithm in WDM rings with wavelength continuity constraint. Moreover, the estimation method allows the calculation of the probability distribution of the constellations of active wavelengths in the network links under the first-fit rule for a given blocking probability.     

A protocol for piggy-backing on Markov based wavelength reservation in WDM optical networks

During lightpath establishment in WDM optical networks, two important steps, other than routing, are: wavelength selection and wavelength reservation. If two or more lightpaths select the same wavelength, wavelength collision occurs. The basic reason for collision is the non-availability of the updated wavelength usage information. Markov based wavelength selection guesses a unique wavelength in advance, where each node periodically broadcasts its adjoining link usage information at a regular interval T. But, at any intermediate time t (0<t<T), all nodes use the old update which is already outdated by time t. If T is large, the problem becomes severe. To get around this problem without introducing any extra overhead, we propose a novel technique where the normal control packets, passing through the relevant nodes during [0,T], can be leveraged to carry the desired link state information. This may allow us even to increase the value of T (thereby reducing broadcast overhead), when control packets are frequent in the network. We apply the proposed modification to our previous work on Markov selection Split Reservation Protocol (MSRP) and call the modified protocol Fast updated MSRP (FMSRP). In fast update, we “piggy-back” normal control packets with link usage information, in addition to the usual periodic broadcasting of the same at every T. It obviously increases the chance for the nodes to get the latest information of link usage. We have simulated FMSRP and compared it with MSRP and also with another current best protocol (i.e., Markov based Backward Reservation Protocol) to show that the blocking probability for FMSRP improves considerably over them in some regions of offered load.