Invalid spectrum rate based scheduling for advance reservation services in elastic optical networks

With the rapid growth of some novel services (e.g. video services), the demand for network bandwidth increases dramatically, which induces an intensive desire in allocating network bandwidth with high flexibility and efficiency. However, the traditional wavelength-division-multiplexing (WDM) optical networks lack mechanism to realize dynamic and efficiency resource allocation. Therefore, elastic optical networks (EONs) have been proposed to accommodate diverse services with heterogeneous network bandwidth and many techniques have been introduced into EONs. Among them, advanced reservation (AR) technique, which has obvious advantage in improving networking performance, has attracted growing interest. Thus, some algorithms employing AR technique have been proposed for EONs. In the traditional slack-based AR algorithm, the starting-time of arriving services can be postponed to release any possible conflicts between the AR services and other existing services, so as to increase the success probability of accommodating the arriving services. However, the traditional algorithm neglected spectrum fragments remained in the network. In this paper, we proposed an Invalid Spectrum Rate (ISR) based scheduling with advance reservation algorithm to optimize the starting-time of arriving services with a comprehensive consideration of both the allocated spectrum and the fragmented spectrum in monitoring the usage of the spectrum before services arrive. In addition, a network load threshold is set on the basis of ISR to trigger the postponement of an arriving request, which help reduce the service blocking probability. The performance of the ISR based scheduling algorithm with different thresholds and different delay times are evaluated by comprehensive simulations and implementations. Experimental results verify that a lower network load and service blocking probability can be achieved by the proposed algorithm compared with the traditional AR algorithm.