ICTON 2011 3RD DAY - Random notes

SESSION We.B3 GOC III

Power Consumption analysis of opaque and transparent optical core network  
A, Autehrieth, A.K. Tilwankar, C. Mas Machuca, J-P. Elbers

Transparent Optical Networks are increasingly more economical over 20Tbps traffic volume. Under that volume opaque network is more economical due to grooming.
The cross point is at an average demand value of 16-30 Gbps in the considered network.


Cross-layer Re-configurable optical network: Fast failure recovery in test-bed for routing algorithm
C. Ware, C.P. Lai, D. Brunina, W. Zhang, A.S. Garg, B. G. Bathula, K. Bergman

Lambda stripping: WDM payload +header wabelengths.
A failuer would trigger the switcher to an alternate port.



The impact of optical networks considering physical impairments, CAPEX and energy consumption
K. Georgakilas, A. Tzanakazi

Wavelength sharing for bkp path with common links assuming the respective primary paths are disjoint.


SESSION We.C3 WAOR III

Dynamic impairment- aware optical networking: Some experimental results of the EU DICONET project
S. Spadaro, J. Perelló, F. Agraz, M. Angelou, S. Azodolmolky, Y. Qin, R. Nejabati, D. Simeonidou, I. Tomkos

A Q-factor is estimated by a Q-tool. A PPD has info on physical impairment on the network
The established connection executed by the RSVP collects information on the physical impairments on the links on the calculated path and this info is to be used by the Q-tool.
It is and enhanced PCE      E-PCE = PCE + NPOT
Centralized light-path establishment.


Off-line algorithms for routing, modulation level and spectrum assignment in elastic optical networks
M. Klinkowski, K. Wakowiak, M. Jaworski

Elastic Optical Networks (EON)
Off-line algorithm. for routing, modulation level and spectrum assignment - RMLSA
Flexible operation: provision of just the resource requirements of a resource
Frequency slots (Jinno et al 2010)
RMLSA problem (Chistodoupoulos et. al 2011)
In a RWA algorithm a whole wavelength is assigned. In a RMSLA problem contiguous slots are assigned.


Experimental implementation of efficient multicast processes: towards carrier Ethernet networks and all-optical multicast
A. Valentí, N. Avallone, A. Pompei, F. Matera, G. Tosi Beleffi

Carrier Ethernet ad all optical multicast
MPLS-TP (transport profile)
Provider backbone bridges
E-Line, E-LAN, and E-tree services
layer 2 multicast by carrier Ethernet network
E-LAN service for multicast traffic live IPTV

Employing Ethernet spanning tree protocols in an integrated hybrid optical network
R. Veisllari, S. Bjørnstad

Hybrid optical network: employs circuit switched path and packet switched path


Optical Labeling through parametric amplification
M.L.F. Abbade, A.L.A. Costa, J.D. Marconi, V.V. Cardoso, H.L. Fragnito, E. Moschim

New optical labeling technology
Label is inserted all optically
Label and payload are emitted a part, in the same BW and different time (strategy for transmission of payload and label)

ICTON 2011 2ND DAY - Random notes

SESSION TuC3 RONEXT III

Survivable Impairment Aware Traffic-grooming and re-generator placement with shared connection-level protection
C. Gao, H.C. Cankaya, A.N. Patel, J.P. Jue, X. Wang, Q. Zhang, P. Palacharla, M. Sekiya

Protection in the connection level x in the light-path level.
Electric backplane: grooming card, line card. Client card.
Transponders (10, 40, 100 Gbps) The cost increases less than the capacity in Gbps.
Reduced signalling, control traffic reduction for hybrid approach of light-path and connection level restoration.

When grooming there is also regeneration since the signal must pass through electronic processing.


Survivable cross-layer virtual topology design using a hyper-heuristic approach
F. Corut Ergin, A Yayimli, A Sima Uyar

An heuristic over a lower level heuristic is used to select which low level heuristic is better to be used.
Ant colony to find a survivable light-path topology.


Cost dependency on protection of optical access networks for dense urban areas
C. Mas Machuca, J. Chen, L. Wosinska

FTTH - PON
Protection in access networks can be highly costly.
Protection until remote node x protection until the user.
cost by deciding previous protection or late reaction.

ICTON 2011 1ST DAY - Random notes

Plenary Speakers

Professor Biswanath Mukherjee
University of California, Davis


Title
Some "Opaque" Problems in Transparent Optical Networks


Trends: Mixed line rates: 20, 40, 100 Gbps wavelengths. (needed for hierarchical grooming)
Elastic Rate networks.
Traffic Engineering: put the traffic where the bandwidth is.
Network engineering: put the bandwidth where the traffic is.
Network planning: put the bandwidth where the traffic is fore-casted to be.


Professor Keren Bergman
Columbia University


Title
Nanophotonic Interconnection Networks for Performance-Energy Optimized Computing


Eletronic x Photonic
Eletronic: Buffer, receive, retransmit at every router
Photonic: larger BW of data. Data is streamed once per communication event.

SESSION Mo.B3

A squatting based framework to enhance network virtualization allocation in optical networks (Invited)
X. Hesselbach, N. Naumenko


Physical resource, virtual resource, logical resource.
Using virtualization helps, for instance the slow implementation of Ipv6 while Ipv4 is still in use. Virtualization helps avoiding the ossification of the Internet.
Classes relate to the amount of capacity or granularity of capacities for instance: lambdas, OBS, fibers, 100Gbps, !0Gbps etc. The demand is per type of service not per resource (BW).
With virtualization resource that is being used in one service can be used in another service.


A GRASP-based heuristic to design the GMPLS control plane network topology with resilience guarantees (Invited)
M. Ruiz, L. Velasco, G. Junyent, J. Comellas


A drawback of a minimal topology is the amount of control information per link is higher. (specially in the case of failure)

Multi layer path computation with PCE part II

From RFC5623

The calculated path is informed to the invoking LSR as a Explicit Route Object and is used for signaling by the RSVP-TE in a MPLS or GMPLS control plane. Two types of computed path may be possible: 

Otpion 1: Mono-Layer Path: 

• Case 1: the lower layer path included in the end-to-end computed path is already established or will be established on demand. This path specified by a ERO contains TE-links may be a regular TE link that is already established or a virtual TE-link that still needs to be established (RFC5212). In the case of a virtual TE link that must be established a setup attempt for a new lower-layer LSP is initiated once the signaling reaches the head-end of the lower-layer LSP. Since the path of a virtual TE-link is not necessarily know in advance a further lower-layer path computation may be needed. 
•  Case 2: the path computed by the PCE contains a loose hop that spans the lower-layer network, it selects which lower-layer to use but does not selects the path across it. A transit Label Switched Router that is the entry for the lower-layer network is responsible for expanding the loose hop in tis own network. The path expansion on the border LSR may rely on the selection of a existing lower-layer LSP or in the computation and set-up of a new one. 

Option 2: Multi-Layer Path:
The path computed by the PCE is a multi-layer path, it contains TE links from diverse layers (RFC4206). This multi-layer path can contain an already established complete path from the lower layer or if the lower layer path is not already established the signaling of the higher layer LSP triggers the establishment of the lower-layer LSP.

Takeda, T.; Oki, E.; Marzin, A.; Farrel, A. & Roux, J. (2009), 'Framework for PCE-Based Inter-Layer MPLS and GMPLS Traffic Engineering', Framework .

Multi layer path computation with PCE

In a distribute multi-layer path computation a Label Switched Router (LSR) in a higher may not have information on the topology or network state of a lower layer and hence may not be able to compute an end-to-end path.
In a centralized multi-layer path computation there are two alternatives for a successful end-to-end path computation:
1- Single PCE path computation: the PCE has topology and network state information about multiple layers in a network and can directly compute a required end-to-end path.
2- Multiple PCE path computation:
 a- with inter PCE communication: PCEs collaborate and each PCE is responsible for the path computation on its own layer.
 b- without inter PCE communication: each PCE computes PCE for its own layer after the request of its layer ingress LSR.

Takeda, T.; Oki, E.; Marzin, A.; Farrel, A. & Roux, J. (2009), 'Framework for PCE-Based Inter-Layer MPLS and GMPLS Traffic Engineering', Framework .