Cost-Minimized Virtual Elastic Optical Network Provisioning with Guaranteed QoS

Network virtualization is important for elastic optical networks (EONs) because of more flexible service provisioning. To ensure guaranteed quality of service (QoS) for each virtual elastic optical network (VEON), clients usually request network resources from a network operator based on their bandwidth requirements predicted from historical traffic demands. However, this may not be efficient as the actual traffic demands of users always fluctuate. To tackle this, we propose a new VEON service provisioning scheme, called SATP, which consists of three stages, i.e., spectrum assignment (SA), spectrum trading (ST), and spectrum purchasing (SP). Unlike conventional once-for-all VEON service provisioning approaches, the SATP scheme first allocates spectrum resources to VEONs according to their predicted bandwidth requirements with a satisfaction ratio alpha (0 < alpha <= 1). Then, to minimize service degradation on VEONs which are short of assigned spectra for their peak traffic periods, the scheme allows VEONs to trade spectra with each other according to their actual bandwidth requirements. Finally, it allows VEON clients to purchase extra spectrum resources from a network operator if the spectrum resources are still insufficient. To optimize this entire process, we for-mulate the problem as a mixed integer linear programming (MILP) model and also develop efficient heuristic algorithms for each stage to handle large test scenarios. Simulations are conducted under different test conditions for both static and dynamic traffic demand scenarios. Results show that the proposed SATP scheme is efficient and can achieve significant performance improvement under both static and dynamic scenarios.

Automated summary

A new VEON service provisioning scheme, SATP, is proposed to address the issue of service quality under dynamic traffic demands by allocating spectrum, allowing spectrum trading, and enabling spectrum purchasing. The scheme is shown to be efficient and achieve significant performance improvement under both static and dynamic scenarios.