Hierarchical Routing And Delay Optimization Algorithms For Optical Virtual Private Networks Using The Hose Model

Optical virtual private networks (OVPNs) is one of the most important services in intelligent optical networks. It not only allows operators to reduce their cost, but also enables customers to control and manage some resources of the private network. Security and stability are provided by OVPN for customers over a shared network. Bandwidth in a link has to be reserved to guarantee the quality of service (QoS). Therefore, how to find a proper path with a least total reserved bandwidth and meeting the demand of QoS has become a hotpot when designing OVPN.This paper mainly studies the routing problem of OVPN in hose model. Hose model specifies upper bounds for both the ingress and egress bandwidth of an endpoint. To minimize the total reserved bandwidth and total average delay, this paper proposed a novel hierarchical routing algorithm. Firstly, it constructs a logical topology among VPN endpoints without considering its underlying physical topology. Secondly, it maps the logical topology into the physical network in terms of different optimal aims. To balance the optimal targets between minimum total reserved bandwidth and minimum total average delay, a customer-inclined parameter is introduced. It enables customers to choose the optimal topology flexibly according to their own requirements. The following conclusions can be derived from numerical results: when customers choose to optimize the total average delay, the total average delay will be reduced by about 30% comparing with optimization of the total reserved bandwidth; Moreover, when choosing any of the two optimal targets, it always requires less total reserved bandwidth than both the primal-dual algorithm and the depth-first spanning tree algorithm. At last, considering the influence of physical parameters in network deployment, this paper optimizes the link loss, and some numerical results are derived to demonstrate the feasibility of the idea.