Abstract

In this paper, a novel Infrastructure as a Service architecture for future Internet enabled by optical network virtualization is proposed. Central to this architecture is a novel virtual optical network (VON) composition mechanism capable of taking physical layer impairments (PLIs) into account. The impact of PLIs on VON composition is investigated based on both analytical model of PLIs and industrial parameters. Furthermore, the impact of network topology on VON composition is evaluated.

© 2011 Optical Society of America

1. Introduction

Future Internet is characterized by global delivery of high-performance network-based applications, such as Cloud Computing and (Ultra) High Definition Video-on-Demand Streaming [1], over a high-capacity dynamic optical network. Due to the requirements for high capacity, low latency, and deterministic quality of service (QoS), dedicated optical network services are desired. However, as these types of applications evolve, the current technical and operational complexities as well as CAPEX and OPEX considerations will limit the ability of network operators to setup and configure dedicated network for each application in a scalable manner [2]. Therefore, a key challenge for network operators is the deployment of dynamic optical infrastructures capable of supporting all application types, each with their own access and network resource usage patterns.

Infrastructure as a Service (IaaS) framework [2] is a key enabler to address this challenge. By adopting IaaS framework and its key technology enabler, Network Virtualization [3], network operators are able to partition their physical infrastructures into virtual networks, based on user/application requirements, and offer them as infrastructure services to users. In that case, multiple virtual networks will coexist over a shared physical infrastructure. These different coexisting virtual networks should be isolated, without interferences between each other. Layer 1 virtual private network (L1VPN) [4] has been proposed as a solution for optical network virtualization. However, the existing L1VPN solutions involve mainly point-to-point dedicated and pre-defined connectivity with deterministic QoS. They are not able to create multiple coexisting virtual networks each with its own network topology where the owners of these virtual networks have full control in terms of routing and management over their portion of the network. Furthermore, optical networks are analogue in nature, which differentiates optical network virtualization from other network virtualization technologies, i.e., layer 2 (L2) and layer 3 (L3) virtualization [3]. Due to the existence of physical layer impairments (PLIs), the adjacent active channels interfere each other, which will impact the isolation of multiple coexisting virtual optical networks (VONs) and the way that VONs are composed. The “PLI-aware” networking is not a new concept, and a lot of work has been done in PLI-aware routing algorithms [5]. However, in optical network virtualization, the impact of PLIs not only on routing paths but also on all the active coexisting virtual networks needs to be investigated.

In this paper, a new IaaS architecture for the future Internet utilizing optical network virtualization is proposed. For this architecture, a novel PLI-aware VON composition mechanism is proposed capable of creating multiple co-existing VONs, each with its own network topology and QoS requirements. The analytical model for assessing the impact of PLIs is also introduced. The impact of PLIs on VON composition is investigated based on both analytical assessment model of PLIs and industrial parameters. Furthermore, the impact of network topology on the VON composition is evaluated. To the authors’ best knowledge, no previous work in the literature has addressed solutions for the network virtualization and optical infrastructure as a service in the context of optical network considering the effect of PLIs.

2. IaaS architecture utilizing PLI-aware optical network virtualization

Virtualization is of paramount importance in next generation networks (NGN) to enable dynamic and efficient infrastructure services [3]. However, virtualization in optical networks entails certain challenges inherent to their analogue nature that do not arise in L2 and L3 virtualization. The main concept of optical network virtualization is to allow the composition of isolated VONs, which are coexisting simultaneously over the same physical infrastructure. A VON, as such, is composed of a set of virtual optical switches (or virtual optical nodes) interconnected by virtual optical links. The virtualization of an optical switch is achieved by partitioning or aggregation of physical optical switches. The associated virtual optical links are defined as the connections between virtual nodes. Therefore, a virtual link may consist of several physical links traversing several physical switches as shown in Fig. 1. The link virtualization granularity (e.g., wavelength or sub-wavelength) depends on the available technology that can guarantee the isolation between virtual links that share the same physical medium. The optical transmission technologies such as DWDM allow the separation of wavelength channels in a single fibre. However, PLIs may heavily affect the isolation between adjacent wavelengths, thus interfering signals that belong to different channels, or in our case, different VONs.

 

Fig. 1 Example of the virtualization of an optical network.

Download Full Size | PPT Slide | PDF

In this paper, a new IaaS architecture enabled by a PLI-aware optical network virtualization mechanism is proposed. The architecture is composed of three layers, as shown in Fig. 2: (i) the optical physical infrastructure layer, which is composed of optical network resources; (ii) the optical virtualization layer, in which the physical resources are virtualized and composed forming VONs; and (iii) the virtual optical network control and management layer that provides control and management functionalities for each VON. The optical virtualization layer is the key innovation and main focus of this architecture. Its functionalities can be simplified in: abstraction to hide technological details and unify the representation of physical resources for virtualization, partitioning/aggregation of nodes and links, and VON composition. VONs are composed by selecting and interconnecting the virtual resources. The functionalities of the control and management layer could be provided by conventional provisioning mechanisms such as GMPLS, so it will stay out of the scope of this paper. Since multiple VONs can coexist sharing the same physical substrate, the interferences caused by the PLIs between VONs will impact the isolation and the performance of each of them. Therefore, the proposed architecture deploys a PLI monitoring and evaluation system within the optical virtualization layer along with a PLI-aware VON composition mechanism. This enables composition and operation of VONs considering the effect of PLIs and guaranteeing a perfect isolation while satisfying the requested QoS.

 

Fig. 2 The reference model of the PLI-aware IaaS architecture.

Download Full Size | PPT Slide | PDF

3. PLI-aware VON composition

The proposed IaaS architecture enabled by optical network virtualization allows service providers to lease resources on demand from infrastructure providers. Moreover, the partitioning and/or aggregation of these resources empowers the creation of multiple simultaneous VONs, each with its own topology and QoS requirements, running over the same optical network infrastructure. Requests for composing VONs are usually generated by service providers or operators. Each request has associated requirements that need to be fulfilled when composing the VON. Given the information of physical network resources and the requirements of VON requests, an intelligent and dynamic composition mechanism is needed to create VONs on demand, utilizing the available physical resources. In this paper, a PLI-aware VON composition mechanism is proposed.

In our work, the physical optical network is modelled as a weighted undirected graph and denoted as Gp = (Np,Ep), where Np is the set of physical optical nodes and Ep is the set of optical links. Each nipNp is associated with the geolocation and the switching capability of the optical node. Each link ep(i, j) ∈ Ep between nodes i and j, with the weight value l(ep) denoting its length, is associated with the number of wavelengths W and the data rate per wavelength R. A VON request is specified as a virtual network topology Gv = (Nv,Ev). The geolocation and attributes of each virtual node nivNv are indicated in the request, as well as the capacity of each virtual link ev(i, j) ∈ Ev, which also shows how virtual nodes are interconnected by virtual links.

Upon the arrival of a VON request, a proper mapping of the VON to the available physical resources is executed by the VON composition mechanism. The proposed PLI-aware VON composition mechanism can be decomposed into three major phases:

  1. Node mapping: Each virtual node nivNv is mapped to a physical node nipNp according to the requirements of geolocation and switching capability of niv.
  2. Link mapping: Each virtual link ev(i, j) ∈ Ev, requesting for certain number of wavelengths, is mapped to a path or a set of physical links ep(i, j) ∈ Ep by adopting routing algorithms (e.g., shortest path or load-balancing routing algorithms). It must be noted that a virtual link connecting two virtual nodes may need to traverse several physical nodes (see Fig. 1). This requires the virtual link mapping process to be able to perform reservation and configuration of such nodes.
  3. Quality verification: The main novelty of the proposed VON composition mechanism is that the PLIs inherent to the optical layer are taken into account, that is, even though there are enough available physical resources to satisfy the requirements of the VON request, the quality of the VON still needs to be verified. Due to the nature of the interferences introduced by PLIs, different active VONs will intervene with each other. Therefore, the quality of all the existing and the newly to be established VONs needs to be verified. For each wavelength wkep(i, j) to be used in the VON, the quality qk(i, j) of all the active wavelengths in the link ep(i, j) will be verified. If the quality of all the involved VONs is acceptable, the wavelength will be allocated to the VON, and the quality of all the involved VONs will be updated. Otherwise, the phase 2 will be executed again till there is no available wavelength resources. The quality verification leads to much more realistic outputs or solutions for service providers to satisfy their users’ requirements.

When a successful VON mapping is achieved, the associated physical network resources are allocated to that VON, ensuring the required level of signal quality. Once the VON is no longer required, all the allocated resources are released.

In our previous work [6], a simple PLI model is used to verify the quality of VONs. In this work, an analytical model for assessing PLIs, also with industrial parameters as inputs, is adopted, which is elaborated in the following section.

4. Analytical model for assessing PLIs

In this study, the nonlinear impairments XPM (Cross-Phase Modulation) and FWM (Four-Wave Mixing) are considered, as well as the linear impairment ASE (Amplified Spontaneous Emission), since they are the major impairments that can impact the signal quality significantly [7].

In order to measure the signal quality, the Q factor is obtained using the following definition,

Q=10lg(Psig_peakσASE2+σXPM2+σFWM2),
where Q is the Q factor, Psig_peak is the peak power of the wavelength channel, σASE2 represents the ASE noise, σXPM2 and σFWM2 represent the non-linear impairments generated by XPM and FWM respectively.

ASE is the noise generated from optical amplifier, which is given as follows,

σASE2=Fhfc(G1)B0,
where F is the EDFA noise figure, h is Planck’s constant, fc is the carrier frequency of a channel under consideration, G is the EDFA optical gain and B0 is the optical bandwidth of the channel.

XPM, as modelled in Eq. (3), is the non-linear phase modulation of an optical channel caused by intensity fluctuations of other co-propagating optical channels, which will be eventually converted to the intensity modulation of the channel [8].

σXPM2=12πjpumps|Hj(w)|2PSDj(w)dw,
where Hj(w) is the XPM transfer function of the pump channel j, and PSDj(w) is the power spectrum of the channel j.

FWM originates from the third order non-linear susceptibility in optical links. When three optical signals in different wavelengths are co-propagating within a fibre simultaneously, a new optical signal is generated in a frequency close to one of the three existing ones, which may interfere with them. The total FWM noise generated on the probe channel is given in Eq. (4).

σFWM2=fi+fjfk=fpi,jkσi,j,k2,
σi,j,k2=PiPjPk|γd31+e2αL2eαLcos(K*L)α2+K2|2,
K=2πλp2c(fifp)(fjfp)[Dλp2c(fi+fj2fp)S],
where σi,j,k2 is the FWM noise generated by one FWM mixing term with the frequency group (fi, fj and fk), and fp is the frequency of the probe channel. If fi, fj or fk equals to fp, Pi, Pj or Pk equals to Psig_peak, otherwise, 12Psig_peak. γ is the nonlinear coefficient of Non-zero dispersion-shifted fibre (NZDSF), d is 3 (degenerate FWM) or 6 (non-degenerate FWM), α is attenuation of NZDSF, L is the length of the optical link, D is the dispersion of NZDSF and S is the dispersion slope of NZDSF.

5. Simulation studies

The simulation studies investigate the impact of PLIs on the VON composition. The PLIs, including ASE, XPM, and FWM, are assessed according to the analytical model elaborated in Section 4. The physical parameters used in the assessment model are given as follows: the data transmission rate per wavelength is 10Gbps; σASE2 of a single EDFA is 10−9 [9]; the channel power Psig_peak is 1mW; the attenuation α, the dispersion D and the dispersion slope S of NZDSF are 0.25dB/Km, 4ps/(Km * nm) and 0.08ps/(Km * nm2) respectively; the nonlinear coefficient of NZDSF is 2/(Km * W). The ITU-T G.694.1 Grid, anchored to 193.1THz (1552.52nm) is adopted and different channel spacing (e.g., 50GHz, 100GHz, and 200GHz) are taken into account in the simulations. The mutual impact of existing VONs and newly requested VON in terms of PLIs is considered by checking the quality of all the active wavelength channels in half of the wavelength window (the wavelength window is the same as the total number of wavelengths). The Q factor threshold is set to a reference value of 8.5dB according to [10].

To investigate the possible issues when facing the challenge of deploying dynamic VON requests, request arrivals are assumed to follow a Poisson process with the average inter-arrival period of 20 time units, and each request has an exponentially distributed holding time with the mean value varying from 100 to 1000 time units [11]. In one instance of simulation, 500 VON composition requests are generated randomly. In each VON request, the number of virtual nodes is randomly determined by a uniform distribution between 2 and half the number of physical network nodes, and each pair of virtual nodes are connected with the probability 0.5. Virtual links are mapped to the physical paths calculated by using the shortest path routing algorithm. For simplicity, only one wavelength is requested for each virtual link. In this study, both scenarios with and without wavelength convertors deployed in core nodes are explored. Three network topologies have been investigated as depicted in Fig. 3.

 

Fig. 3 Physical topology with average node degree shown in paranthesis.

Download Full Size | PPT Slide | PDF

5.1. The impact of wavelength channel spacing on VON composition

We first evaluate the impact of different wavelength channel spacing values on the VON composition. The failure rate of VON composition is collected as the performance comparison criteria, which is defined as the ratio of the number of declined VON requests relative to the total number of VON requests. In this simulation, the NSF topology with 14 nodes-21 links is adopted, and the number of wavelengths per link is set to 16. The channel spacing is ranged from 50GHz to 200GHz. As the channel spacing increases, the interference between adjacent channels caused by PLIs (XPM and FWM) is reduced. Therefore, two VONs are more likely to use two adjacent wavelength channels and wavelength resources are used more efficiently. This is reflected in Fig. 4, where results verify the analytical model for assessing PLIs and the established simulation platform. As expected, results show that when virtualizing optical networks, more VONs can be composed over the same optical physical substrate by increasing the channel spacing. On the other hand, the figure also depicts that for longer VON holding time, requests are blocked with a higher probability.

 

Fig. 4 The impact of different channel spacing on the VON composition.

Download Full Size | PPT Slide | PDF

5.2. The performance evaluation of PLI-aware VON composition

In order to evaluate the PLI-aware VON composition mechanism, two different scenarios have been analyzed with and without Q verification (i.e., “Q Verification” and “no Q Verification” in Fig. 5 and 6). In the scenario with “no Q Verification”, VON requests are composed without considering the impact of PLIs. On the other hand, in the scenario with “Q Verification”, a VON request is composed only using the available physical resources that can meet the acceptable Q factor threshold (i.e., 8.5dB). From the results, we can observe the severe impact of PLIs on the VON composition. The quality verification rejects almost half of the VON requests (about 54.72% when the number of wavelengths per link is 16, the VON holding time is 400 units, and the channel spacing is 50GHz) due to the unacceptable quality despite the availability of spare resources.

 

Fig. 5 The impact of PLIs on the VON composition with different number of wavelengths per link. VON holding time is 400 units. Channel spacing is 50 GHz.

Download Full Size | PPT Slide | PDF

 

Fig. 6 The impact of PLIs on the VON composition with different VON holding times.

Download Full Size | PPT Slide | PDF

Results in Fig. 5(a) also show that, as the number of wavelengths per link increases, the failure rate of VON composition with Q verification remains almost the same. This is due to the fact that wavelength selection uses the First-Fit (FF) algorithm. However, the chosen wavelengths are always adjacent to the channels that are currently being used, and most of them have unacceptable quality due to PLIs. Moreover, in this scenario, the quality verification of VON is executed after the wavelength selection, and the quality verification does not give any feedback to the resource checking process. Therefore, after the quality verification failure, the VON request is directly rejected. In this case, even if the number of wavelengths per link increases, the failure rate of VON composition will not change significantly. To compensate this effect, in the proposed PLI-aware VON composition algorithm, wavelength selection is just performed over an already verified wavelength set that can guarantee the required quality. The results of this mechanism, named PLI-aware FFWA, are also shown in Fig. 5 and 6, and compared with the two previous scenarios. The results indicate that the proposed algorithm can improve the performance of VON composition (about 65.95% when the number of wavelengths per link is 16, the VON holding time is 400 units, and the channel spacing is 50GHz). The impact of the wavelength conversion (WC) capability is also evaluated, and the results are shown in Fig. 5(b). Without wavelength converters being installed, all the virtual links have to be mapped to the same wavelength to maintain the connectivity of an entire VON, which will suffer the limit of physical resources severely, as reflected in the results.

5.3. The impact of node degree on PLI-aware VON composition

The impact of network topology, i.e., node degree, on the PLI-aware VON composition has also been evaluated in the six-node simple and full mesh topologies shown in Fig. 3. It is assumed that all physical links in the two topologies have the same length and each wavelength channel has the same data rate. Moreover, to make a fair comparison over the two topologies the total number of resources is the same (120 wavelengths), that is, 8 wavelengths in the full mesh network topology (15 links in total) and 15 wavelengths in the simple network topology (8 links in total). In Fig. 7, the results show that the optical network with a higher node degree can achieve better VON composition performance (about 32.43% higher when the total number of wavelengths in each network is 120, the VON holding time is 400 units, and the channel spacing is 50GHz). The improvement is due to the fact that in a full mesh network, virtual links can be directly mapped into physical links one to one, whilst with lower node degree a virtual link may need to traverse multiple physical links experiencing more severely the effect of PLIs.

 

Fig. 7 The impact of node degree on the PLI-aware VON composition with different VON holding time. Channel spacing is 50 GHz.

Download Full Size | PPT Slide | PDF

6. Conclusions

This paper proposed a novel IaaS architecture for the future Internet exploiting PLI-aware optical network virtualization. For this architecture, a new impairment-aware VON composition mechanism is proposed. By adopting analytical model of PLIs as well as industrial parameters, network simulation studies show that the proposed mechanism can mitigate the impact of PLIs on the VON composition. The impact of the wavelength channel spacing and the wavelength conversion capability on the VON composition is also evaluated. Furthermore, the results also show that the PLI-aware VON composition mechanism performs better in the network with higher node degree. The presented results can provide infrastructure providers with guidelines for properly planning their networks to serve VON requests with acceptable quality.

Acknowledgments

The work described in this paper has been carried out with the support of the EU funded FP7 GEYSERS project and UK funded EPSRC PATRON project.

References and links

1. Cisco white paper, “Cisco Visual Networking Index: Forecast and Methodology, 2009–2014”.

2. S. Figuerola and M. Lemay, “Infrastructure Services for Optical Networks (Invited),” J. Opt. Commun. Netw. 1(2), A247–A257 (2009). [CrossRef]  

3. M. Chowdhury and R. Boutaba, “A Survey of Network Virtualization,” Comput. Netw. 54(5), 862–876 (2010). [CrossRef]  

4. T. Takeda, “Framework and Requirements for Layer 1 Virtual Private Networks,” RFC 4847 (2007).

5. C. V. Saradhi and S. Subramaniam, “Physical Layer Impairment Aware Routing (PLIAR) in WDM Optcial Networks: Issues and Challenges,” IEEE Commun. Surv. Tutor. 11(4), 109–130 (2009). [CrossRef]  

6. S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).

7. G. P. Agrawal, Fiber-Optic Communication Systems, 3rd. ed. (Wiley-Interscience, 2002). [CrossRef]  

8. W. Lin, “Physically Aware Agile Optical Networks,” Ph.D. dissertation (Montana State University-Bozeman, 2008).

9. W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011). [CrossRef]  

10. G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009). [CrossRef]  

11. X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011). [CrossRef]  

References

  • View by:
  • |
  • |
  • |

  1. Cisco white paper, “Cisco Visual Networking Index: Forecast and Methodology, 2009–2014”.
  2. S. Figuerola and M. Lemay, “Infrastructure Services for Optical Networks (Invited),” J. Opt. Commun. Netw. 1(2), A247–A257 (2009).
    [Crossref]
  3. M. Chowdhury and R. Boutaba, “A Survey of Network Virtualization,” Comput. Netw. 54(5), 862–876 (2010).
    [Crossref]
  4. T. Takeda, “Framework and Requirements for Layer 1 Virtual Private Networks,” RFC 4847 (2007).
  5. C. V. Saradhi and S. Subramaniam, “Physical Layer Impairment Aware Routing (PLIAR) in WDM Optcial Networks: Issues and Challenges,” IEEE Commun. Surv. Tutor. 11(4), 109–130 (2009).
    [Crossref]
  6. S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).
  7. G. P. Agrawal, Fiber-Optic Communication Systems, 3rd. ed. (Wiley-Interscience, 2002).
    [Crossref]
  8. W. Lin, “Physically Aware Agile Optical Networks,” Ph.D. dissertation (Montana State University-Bozeman, 2008).
  9. W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011).
    [Crossref]
  10. G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
    [Crossref]
  11. X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
    [Crossref]

2011 (2)

W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011).
[Crossref]

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

2010 (1)

M. Chowdhury and R. Boutaba, “A Survey of Network Virtualization,” Comput. Netw. 54(5), 862–876 (2010).
[Crossref]

2009 (3)

C. V. Saradhi and S. Subramaniam, “Physical Layer Impairment Aware Routing (PLIAR) in WDM Optcial Networks: Issues and Challenges,” IEEE Commun. Surv. Tutor. 11(4), 109–130 (2009).
[Crossref]

G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
[Crossref]

S. Figuerola and M. Lemay, “Infrastructure Services for Optical Networks (Invited),” J. Opt. Commun. Netw. 1(2), A247–A257 (2009).
[Crossref]

Agrawal, G. P.

G. P. Agrawal, Fiber-Optic Communication Systems, 3rd. ed. (Wiley-Interscience, 2002).
[Crossref]

Antoniades, N.

G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
[Crossref]

Azodolmolky, S.

S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).

Boutaba, R.

M. Chowdhury and R. Boutaba, “A Survey of Network Virtualization,” Comput. Netw. 54(5), 862–876 (2010).
[Crossref]

Cheng, X.

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

Chowdhury, M.

M. Chowdhury and R. Boutaba, “A Survey of Network Virtualization,” Comput. Netw. 54(5), 862–876 (2010).
[Crossref]

Ellinas, G.

G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
[Crossref]

Escalona, E.

S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).

Figuerola, S.

Hadjiantonis, A.

G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
[Crossref]

Hahn, T.

W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011).
[Crossref]

Lemay, M.

Levine, A. M.

G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
[Crossref]

Lin, W.

W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011).
[Crossref]

W. Lin, “Physically Aware Agile Optical Networks,” Ph.D. dissertation (Montana State University-Bozeman, 2008).

Luo, Y.

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

Mumey, B.

W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011).
[Crossref]

Nejabati, R.

S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).

Panayiotou, T.

G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
[Crossref]

Peng, S.

S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).

Saradhi, C. V.

C. V. Saradhi and S. Subramaniam, “Physical Layer Impairment Aware Routing (PLIAR) in WDM Optcial Networks: Issues and Challenges,” IEEE Commun. Surv. Tutor. 11(4), 109–130 (2009).
[Crossref]

Simeonidou, D.

S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).

Su, S.

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

Subramaniam, S.

C. V. Saradhi and S. Subramaniam, “Physical Layer Impairment Aware Routing (PLIAR) in WDM Optcial Networks: Issues and Challenges,” IEEE Commun. Surv. Tutor. 11(4), 109–130 (2009).
[Crossref]

Takeda, T.

T. Takeda, “Framework and Requirements for Layer 1 Virtual Private Networks,” RFC 4847 (2007).

Wang, H.

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

Wang, J.

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

Wolff, R. S.

W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011).
[Crossref]

Yang, F.

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

Zhang, Z.

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

ACM SIGCOMM Comput. Commun. Rev. (1)

X. Cheng, S. Su, Z. Zhang, H. Wang, F. Yang, Y. Luo, and J. Wang, “Virtual network embedding through topology-aware node ranking,” ACM SIGCOMM Comput. Commun. Rev. 41(2), 38–47 (2011).
[Crossref]

Comput. Netw. (1)

M. Chowdhury and R. Boutaba, “A Survey of Network Virtualization,” Comput. Netw. 54(5), 862–876 (2010).
[Crossref]

IEEE Commun. Surv. Tutor. (1)

C. V. Saradhi and S. Subramaniam, “Physical Layer Impairment Aware Routing (PLIAR) in WDM Optcial Networks: Issues and Challenges,” IEEE Commun. Surv. Tutor. 11(4), 109–130 (2009).
[Crossref]

IEEE Photon. Technol. Lett. (1)

G. Ellinas, N. Antoniades, T. Panayiotou, A. Hadjiantonis, and A. M. Levine, “Multicast Routing Algorithms Based on Q-Factor Physical-Layer Constraints in Metro Networks,” IEEE Photon. Technol. Lett. 21(6), 365–367 (2009).
[Crossref]

J. Opt. Commun. Netw. (1)

Opt. Switching Networking (1)

W. Lin, T. Hahn, R. S. Wolff, and B. Mumey, “A distributed impairment aware QoS framework for all-optical networks,” Opt. Switching Networking 8(1), 56–67 (2011).
[Crossref]

Other (5)

Cisco white paper, “Cisco Visual Networking Index: Forecast and Methodology, 2009–2014”.

T. Takeda, “Framework and Requirements for Layer 1 Virtual Private Networks,” RFC 4847 (2007).

S. Peng, R. Nejabati, S. Azodolmolky, E. Escalona, and D. Simeonidou, “An Impairment-aware Virtual Optical Network Composition Mechanism for Future Internet,” in Proceedings of ECOC, Tu.6.K.3 (2011).

G. P. Agrawal, Fiber-Optic Communication Systems, 3rd. ed. (Wiley-Interscience, 2002).
[Crossref]

W. Lin, “Physically Aware Agile Optical Networks,” Ph.D. dissertation (Montana State University-Bozeman, 2008).

Cited By

OSA participates in Crossref's Cited-By Linking service. Citing articles from OSA journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (7)

Fig. 1
Fig. 1 Example of the virtualization of an optical network.
Fig. 2
Fig. 2 The reference model of the PLI-aware IaaS architecture.
Fig. 3
Fig. 3 Physical topology with average node degree shown in paranthesis.
Fig. 4
Fig. 4 The impact of different channel spacing on the VON composition.
Fig. 5
Fig. 5 The impact of PLIs on the VON composition with different number of wavelengths per link. VON holding time is 400 units. Channel spacing is 50 GHz.
Fig. 6
Fig. 6 The impact of PLIs on the VON composition with different VON holding times.
Fig. 7
Fig. 7 The impact of node degree on the PLI-aware VON composition with different VON holding time. Channel spacing is 50 GHz.

Equations (6)

Equations on this page are rendered with MathJax. Learn more.

Q = 10 lg ( P sig _ peak σ A S E 2 + σ X P M 2 + σ F W M 2 ) ,
σ A S E 2 = F h f c ( G 1 ) B 0 ,
σ X P M 2 = 1 2 π j pumps | H j ( w ) | 2 P S D j ( w ) d w ,
σ F W M 2 = f i + f j f k = f p i , j k σ i , j , k 2 ,
σ i , j , k 2 = P i P j P k | γ d 3 1 + e 2 α L 2 e α L cos ( K * L ) α 2 + K 2 | 2 ,
K = 2 π λ p 2 c ( f i f p ) ( f j f p ) [ D λ p 2 c ( f i + f j 2 f p ) S ] ,

Metrics