Delay tolerance aware queueing and scheduling for differentiated services in dynamic survivable WDM networks

Set-up delay tolerance is an important metric that can be used in user-controlled and bandwidth intensive applications such as IPTV, backup storage, and grid computing. It is defined by customer and stated in service level agreement (SLA). Therefore, service differentiation based on the delay tolerance and holding time provides great opportunity for service providers to maximize their profit. The service classes with high delay tolerance could have been serviced with some delay. However, they are traditionally provisioned without tolerating any delay. In this study, we use the set-up delay tolerance more efficiently. We intentionally insert these service classes into a queue to save optical resources for future requests. Moreover, in order to increase the acceptance rate of the waiting connections, we develop a strategy rescheduling the queue after each departure event and chooses the best group of the waiting connections to provision. Different algorithms, Dif-SC-2-3 and Dif-SC-3-G, are presented in this paper based on these approaches. For a typical wavelength convertible US nationwide and Pan-European network (USNET and EON), our approach results in significant decrease in overall blocking probability (BP), while reducing the BP of the service classes with stringent delay tolerance as well. For example, around 100 Erlangs, Dif-SC-3-G algorithm provides a BP reduction of up to 72% and 75% in USNET and EON, respectively. In addition, the proposed algorithms improve the fairness. Around 100 Erlangs, Jain׳s fairness indexes for Dif-SC-2-3 are the best values which are 97% and 96% in USNET and EON, respectively.