Yining Cao

Dynamic priority-based alternate routing for multiple classes of traffic in intelligent optical networks

Priority-based alternate routing (PAR) and dynamic priority-based alternate routing (DPAR) are introduced for multiple classes of traffic in intelligent optical networks. A discrete event simulation platform was implemented to compare the performance of different priority schemes. As usual, the call interarrival and holding time for each class of traffic were assumed to be exponentially distributed in the simulation. The results show that PAR can significantly decrease the blocking probability of optical networks, especially for traffic with low priority. Because the weights of links of the network in DPAR increase with the increment of wavelength occupation, links with more idle wavelengths are more easily chosen as part of the shortest path. This mechanism benefits to the load balance of the network and as a result, DPAR has better performance than PAR. In order to utilize the advantage of the wavelength reservation-based priority scheme (RES), which provides low blocking for traffic with high priority, the combination of DPAR and RES was also investigated. The simulation results show that DPAR with the proper RES provides the lowest blocking probability for every class of service compared to the corresponding class of other priority schemes.

Differentiated Service Provisioning Based on Lightpath Length Estimation in overlay IP-over-WDM Networks

A strategy is proposed to handle traffic of different priorities in overlay IP-over-WDM networks, including a lightpath length estimation scheme and a QoS admission control mechanism. Simulation results show it effectively provides differentiated services.

Adaptive Wavelength Assignment Scheme for Distributed Path Restoration in Optical Networks

In wavelength-routed optical networks, routing and wavelength assignment (RWA) can be implemented in a dynamic and distributed way. During the distributed restoration process after a failure occurs, the concurrent reservation attempts from different original nodes may lead to collisions because they select the same wavelength with a large possibility, resulting in high recovery blocking probabilities. This paper presents a contention avoidance mechanism to improve restoration performance. With the adaptive wavelength assignment method, reservation contentions between adjacent nodes are reduced significantly and higher recovery ratio is achieved.

Hop constraint-based capacity fairness control in IP-over-WDM networks

In this article, we consider the capacity fairness problem in IP-over-WDM networks. Since connections with different bandwidth granularities may be established over a shared lightpath, fairness in bandwidth allocation among different users becomes a crucial problem. A simple, yet efficient hop constraint-based admission control scheme is proposed to accommodate more high-bandwidth requests. Through rejecting some of the low-capacity requests that would go through alternative paths with more hops and thus would consume a larger amount of bandwidth, the blocking probability of high-capacity requests reduces notably. Numerical results show that this proposal achieves significant improvement in capacity fairness without raising the overall blocking probability. In addition, it achieves excellent fairness performance at both light and heavy loads by selecting the rejection probability dynamically.

Hop constraint based admission and fairness control in IP-over-WDM networks

From both user and operator perspectives, fairness is an important aspect in IP-over-WDM networks where Label Switched Paths (LSPs) are dynamically groomed over optical networks. The setup of LSPs with long distances experiences a higher blocking probability due to both lightpath establishment unfairness in the optical layer and link cascading effect in the IP/MPLS layer. A simple LSP connection admission and fairness control mechanism is proposed in this article. This control mechanism is based on hop constraint, in which an LSP is accepted with a pre-assigned probability according to its distance and the hops of its route. Through suppressing connection of short-distance LSPs that overuse bandwidth to facilitate the setup of LSPs with long distances, this proposal achieves fine distance fairness performance with a slight overall blocking probability increment.

Collision avoidance wavelength assignment scheme for distributed path restoration in optical networks

In wavelength-routed optical networks, the recovery ratio of interrupted connections after failures is mainly affected by adjacent-node resource reservation collisions in a distributed path restoration. This paper proposes a novel collision avoidance wavelength assignment (CA-WA) mechanism to make adjacent nodes select different wavelength assignment schemes and therefore different recovery wavelengths for the recovery routes. Simulation results show that in both single- and double-fiber failure scenarios the proposed wavelength assignment mechanism significantly reduces reservation contentions and achieves a higher recovery ratio, especially when the network load is high.

Combined analysis of cost and traffic grooming policies for hybrid networks under dynamic traffic requests

The benefit of a two-layer hybrid IP/MPLS (multi-protocol label switching) over a wavelength division multiplexing network has been analyzed considering both the cost and different grooming policies. A detailed cost and performance analysis of hybrid networks is done for three different grooming policies. The hybrid network cost is compared with that of an opaque network for equal traffic demand and equal blocking probability of dynamic requests of label switched paths. An algorithm is given to design optimum hybrid nodes for different grooming policies to provide the desired blocking probability for a given number of dynamic connection requests. The results show that all three applied grooming policies (IP layer first, optical layer first, and one hop first) result in lower costs of the hybrid network architecture than for the opaque network. In addition, an adaptive one hop first method is given to improve the best of the applied grooming policies, which limits grooming in heavily loaded hybrid nodes to achieve load balancing. The simulation results show that the new policy significantly reduces the overall blocking probability.

Overlap-extend hierarchical restoration for Metropolitan Area Network

The restoration time is an important network survivability index in large-scale Metropolitan Area Network. The early study indicates that the hierarchical restoration mechanism have lower restoration time, comparing with global restoration. But the restoration time increases rapidly when the restoration scope expands. In this paper, Overlap-extend hierarchical restoration mechanism is proposed to reduce the restoration time, by restricting the restoration in a definite scope. A theory model is established to describe the average restoration time of Overlap-extend mechanism and existing hierarchical restoration mechanism, which can be used to analyse the influencing factors in our mechanism. Simulation result is also presented, which shows that overlap-extend mechanism can observably reduce restoration time, especially in light traffic load.