SDH Layered Architecture

8/8/2019 SDH Layered Architecture

1/19

Feature Articles: Telecom Operation & Management

China Communications August200572

ABSTRACT

This paper discusses the evolution path for core and

metropolitan networks taking into account the cur-

rent economic recovery as well as the changing tele-

communications environment. At the beginning of

the paper, the current status of core and metropoli-

tan networks is presented, including a brief presen-

tation of such networking technologies as: MPLS,

Ethernet, Resilient Packet Ring (RPR), SDH/

SONET, and OTN. Then, the evolution scenarios

are provided in three stages: short, medium and long

term. The following factors are taken into account

and referred to at each step of the evolution scenario:

available services, quality of service/traffic

engineering, connection provisioning/connection

set-up control methods, network resilience and other

functions. The short term scenario involves the in-

Planning of Optical Transport

Networks Layered Architecture

A. Jajszczyk A. Lason, J. Rzasa, R. Stankiewicz

AGH University of Science and Technology, Department of Telecommunications,

Al. Mickiewicza 30, 30-059 Krakow, Poland

e-mail: {jajszczyk, lason, rzasa, rstankie}@kt.agh.edu.pl

M. Jaeger

T-Systems, Goslarer Ufer 35, 10589 Berlin, Germany

e-mail: [email protected]

S. Spadaro, J. Sole-Pareta

Universitat Politecnica de Catalunya (UPC), C/ Jordi Girona 1-3, 08034 Barcelona, Spain

e-mail: [email protected], [email protected]

troduction of reconfigurable WDM networks, while

in the medium term the Generic Framing Procedure

(GFP), enhanced SDH/SONET technologies, and

Optical Transport Network (OTN) will be adopted.

The long term scenario deals with the addition of a

control plane, either ASON or GMPLS based.

Key words: OTN, SDH/SONET, ethernet, RPR,

IP, network planning, ASON/GMPLS, GFP, LCAS,

DWDM/CWDM

INTRODUCTION

The late 90s were boom times in the telecommuni-

cations industry. Thousands of kilometers of fiber

were installed, many transport systems implemented,

dozens of new companies, both operators and vendors,

started their fight for profit. However, the subsequent

slowdown in the world economy has brought some

FEATURE ARTICLESFEATURE ARTICLESFEATURE ARTICLESFEATURE ARTICLESFEATURE ARTICLES


2/19


China Communications August2005 73

companies at the edge of catastrophe. Recently, the

mood seems to start not to be so pessimistic. However,

the question how to develop the most effective and

cost efficient telecommunications infrastructure is of

utmost importance. At the beginning of the paper, wepresent the architecture of current core and metro-

politan networks and point out their main drawbacks.

Next, we propose an evolution path which meets the

increasing requirements for performance, function-

ality and cost efficiency, and is applicable to both core

and metropolitan network operators. The paper shows

a feasible way to solve difficult task of building an

evolution path for transport networks. The path pro-

posed here, presents one of possible ways which the

core and metropolitan networks may follow.Certainly, depending on the type of networks, busi-

ness model, technical constraints, etc., a different evo-

lution path can be drawn. For example, some tech-

nologies may be implemented faster by a newcomer

than by an incumbent core operator while metropoli-

tan carriers can omit such technology at all. However,

our evolution path seems to be representative for cur-

rent core and metropolitan networks. During the pro-

cess of building the evolution path, a wide range of

issues has to be taken into account, ranging fromtechnical, economic, organizational to social ones.

Nevertheless, our paper is mainly focused on techni-

cal and, in some parts, economic issues.

The following factors are taken into account at

each step of the evolution scenario:

services, e.g.,Bandwidth on Demand Service,

Provisioned Bandwidth Service and Optical Vir

tual Private Network,

quality of service/traffic engineering,

connection provisioning/connection set-up con

trol methods (permanent connections (PC),

soft-permanent connections (SPC), switched

connections (SC)),

resilience functions, such as protection and

restoration,

control plane functions, such as routing and

signaling,

drivers behind each step, etc.,

All these factors will allow us to evaluate current

as well as future steps of the evolution path in a way

which permits network operators to offer services to

customers. Provisioned Bandwidth Service (PBS)

denotes here static near-real-time provisioning through

management interfaces via a network management

system (NMS) or an operations support system (OSS)

with a client-server relationship between clients and

the optical network. In contrast,Bandwidth on De-

mand Service (BDS) denotes dynamic and real-time

provisioning in seconds or sub-seconds with signaled

connection requests via a User to Network Interface

(UNI). Optical Virtual Private Network(OVPN)

specifies a set of provided network resources, e.g.,

link bandwidth, wavelength, and/or optical connec-

tion ports that may be used. For clients belonging toan OVPN, a Closed User Group (CUG) and a virtual

network are defined, where optical connections may

be based on static or dynamic (signaled) provisioning.

The resource visibility and its control vary depend-

ing on the service contract.

The organization of the remainder of this paper is

as follows. We start with a description of the current

status of transport and metropolitan networks. Then,

an evolution scenario for a short term time scale is

provided. Subsequently, medium and long term sce-narios are presented.

Throughout the article, the authors refer to the stan-

dardization process which is carried out by theIn-

ternational Telecommunication Union Telecommu-

nication Standardization Sector(ITU-T), theInternet

Engineering Task Force (IETF) as well as theInsti-

tute of Electrical and Electronics Engineers (IEEE).

Moreover, the authors refer to other bodies, e.g., the

Optical Internetworking Forum (OIF), theMetro

Ethernet Forum (MEF) and theResilient Packet RingAlliance (RPRA).

CURRENT STATUS

OF TRANSPORT NETWORKS

We start this chapter with an overview of main net-

working technologies used in modern transport

networks. Next, we present the layered structure of


3/19



transport networks and outline their crucial

characteristics.

1. Wavelength Division Multiplexing

Todays transport networks widely use the wave-

length division multiplexing(WDM), i.e., circuit

switching technology and, in some cases, single-

wavelength optical fibers, referred to as Tradi-

tional Fiber Optics (TFO). Wavelength division

multiplexing is deployed for point-to-point com-

munications with manually configured links. The

spectacular increase of the capacity aggregated

by the fiber optics removed the bandwidth bottle-

neck in the core, regional and metropolitan

networks. WDM systems already installed by

some network operators offer up to 40 Gb/s data

rate per optical carrier and 160 carriers per fiber.

The ITU-T recommendations specify a broad

range of aspects related to optical networks

including, e.g., the physical layer and WDM.

Recommendations on the fiber optics physical

layer are already in place (G.65x series of

Recommendations), coarse WDM (CWDM)

wavelength and dense WDM (DWDM) fre-

quency grids are available as well (G.694.2 andG.694.1, respectively) [1], [2]. The recent advances

in optical layer technology enable the architec-

ture optimization of telecommunication transport

networks. Currently, the main effort of equip-

ment vendors and network operators concerns the

architectural aspects of the optical layer. The in-

troduction of optical cross-connects OXC and op-

tical add-drop multiplexers (OADM) are ex-

pected to lead to major cost reductions in over-

all networking due to reduced electronic signal

processing and limited use of expensive opto-

electronic conversion.

Besides WDM, single-wavelength fiber optics

systems are in use as well, however, usually at much

shorter distances (less than 50 km). Hence, the TFO

is typically applicable to some metropolitan

networks. Nevertheless, it is broadly believed that

the number of traditional fiber optics systems in

transport networks will decrease.

2. SDH/SONET and Virtual Concatenation

On top of WDM or TFO, the Synchronous Digital

Hierarchy (SDH) or Synchronous Optical Network

(SONET) is extensively used. The SDH/SONETis a circuit switching technology and is applicable

to both metropolitan and core networks. It is well-

understood, mature and standardized [3], [4], [5]. Since

it was initially designed to optimize transport of

64-kb/s-based TDM services, a rigid capacity of

payload as well as a coarse fixed-rate multiplexing

hierarchy was defined. Today, SDH/SONET sys-

tems are built with bit rates as high as 10 Gb/s

(STM-64/OC-192), with 40 Gb/s (STM-256/OC-

768) on the horizon. Current SDH/SONET core

networks have a switching granularity of VC-4/

STS-3. A majority of all client networks are set up

on top of SDH/SONET.

By the use of Virtual Concatenation (VC)

procedure, SDH/SONET may be improved to better

meet todays requirements, e.g., various switching

granularities. Virtual Concatenation [3] allows flex-

ible concatenation of several SDH/SONET payloads.

It assures an effective use of SDH/SONET capacity.

Virtually concatenated payloads constitute a Virtual

Concatenation Group (VCG). Members of a VCG,

as opposed to contiguous concatenation, may not

reside in the same STM-N/OC-Ncontiguously. They

may even reside at different STM-N/OC-Ninterfaces

and are treated within the network separately and

independently. As a consequence, members of a

VCG may reach the destination through various

routes. Intermediate nodes do not need to handle

virtual concatenation. The VC functionality must be

implemented only at path termination nodes. This

feature makes it possible to deploy virtual concat-

enation on legacy SDH/SONET equipment of ex-

isting networks, thus to smooth transition to en-

hanced networks. On the other hand, it should be

noted that differences in the delay of an individual

concatenated signal may occur due to pointer pro-

cessing at intermediate nodes. Compensation of dif-

ferential delays is handled at the destination node.

Another advantage of virtual concatenation is its


4/19



ability to divide STM-N/OC-Nbandwidth into sev-

eral subrates. Each of the subrates may be used for

accommodation of a different service. The band-

width of STM-N/OC-Nmay be shared, for example,

by both telephone service and data signals. An of-

ten-mentioned example [6] of a practical use of vir-

tual concatenation is Gigabit Ethernet. VC-4-16c

(STM-16) is required to accommodate Gigabit

Ethernet signals at full speed under conventional

SDH. However, the capacity of 1.4 Gb/s is then

wasted. On the other hand, contiguous concatena-

tion of four VC-4 containers (VC-4-4c) provides too

small capacity to fully accommodate Gigabit

Ethernet signals. The best solution would be con-

catenation of seven VC-4 payloads. It is possiblewith virtual concatenation. Bandwidth of 1.05 Gb/s

provided by a VC-4-7v VCG is suitable for Gigabit

Ethernet. More examples of bandwidth efficiency

in carrying Ethernet, Fast Ethernet and Gigabit

Ethernet data signals in SDH with and without VC

are shown in Table 1.

3. Ethernet

Ethernet networks are important clients of the trans-

port layer. We use the termEthernetin the mean-

ing of traditionalEthernet, Fast Ethernet, Gigabit

Ethernetas well as 10 Gigabit Ethernet[7], [8]. The

Ethernet technology is well understood and robust,

its applicability to local computer networks cannot

be questioned. Since years, 10 and 100 Mb/s

Ethernets have been used for building cost effective,

high speed data networks. In recent years, Gigabit

Ethernet widely found its way into the metropolitan,

regional and even wide area networks. 10 Gigabit

Ethernet continues the evolution towards higher bit

rates and an extended range, although, data are

transferred by fiber links only. In this case, two

types of physical interfaces were defined, the first

one is suitable for local and metro area networks

operation (LAN PHY: 10GBase-X, 10GBase-R),

the second for wide area networks (WAN PHY:

10GBase-W). The 10 Gigabit Ethernet standard

proposes physical interfaces based both on single-

and multi-mode fibers. 10 Gigabit Ethernet LAN

PHY offers an extended reach compared to Giga-

bit Ethernet, i.e., over a 40 km long single-mode

fiber link. WAN PHY differs from the LAN PHY

implementation by the use of the SDH/SONET

framing with reduced functionality. The framing

for WAN interfaces takes place at the WAN Inter-

face Sublayer(WIS). The output from the WANPHY is compatible with the synchronous frame

format (VC-4-64c or STS-192c) and can be easily

transported over an Optical Transport Network

(OTN). The output from the LAN PHY interface

of 10 Gigabit Ethernet has to be adapted before

entering the OTN. The newly proposed Generic

Framing Procedure format promises to provide this

function. The Ethernet technology was also pro-

posed as the base for new high speed access

networks. TheEthernet in the First Mile working

group, 802.3ah, was formed within the IEEE 802.

3 CSMA/CD working group. The scope of the work

is the adaptation of the Ethernet technology to

point-to-point and point-to-multipoint (E-PON)

access networks [9]. A successful standardization

process will extend the Ethernet coverage so that

end-to-end Ethernet services can be offered to both

business and residential customers. Future improve-

ment of the quality of service functions offered in

Ethernet networks can be achieved through the use

of 802.1D (Class of Services) and 802.1Q (Virtual

payload mapping bandwidth efficiency payload mapping bandwidth efficiencyEthernet (10 Mb/s) VC-3 21% VC-11-7v 89%

Fast Ethernet (100 Mb/s) VC-4 67% VC-11-64v 98%

Gigabit Ethernet (1 Gb/s) VC-4-16c 42% VC-4-7v 95%

Data signalSDH without VC SDH with VC

Table 1 Bandwidth efficiency of virtual concatenation


5/19



LAN) specifications. Unlike SDH/SONET, the

Ethernet technology does not provide any fast pro-

tection mechanism. Ethernet generally relies on the

spanning tree protocol to eliminate all loops from

a switched network. Even though the spanning tree

protocol can be used to achieve path redundancy,

it recovers comparatively slowly from a fiber cut,

as the recovery mechanism requires the failure con-

dition to be propagated serially to each upstream

node. IEEE 802.1DRapid Spanning Tree Protocol

(RSTP) improves resiliency of the Ethernet [10].

However, SDH/SONET-like services still cannot

be guaranteed, hence, the Ethernet suffers from in-

ability to provide carrier class services. Although,

some early works have been done by the MetroEthernet Forum in its specification [11], it seems that

it is still too early to fully introduce Ethernet based

carrier-class services in the metro. Metro Service

Model Phase 1 proposes service building blocks or

service attributes and specifies how to build an

Ethernet service. Such services, described as

Ethernet Line, i.e., point-to-point services and

Ethernet LAN, i.e., multipoint-to-multipoint

service, may be offered over fiber, SDH/SONET

or WDM technology.

4. Resilient Packet Ring

Resilient Packet Ring is a new technology for ring-

based metropolitan area networks that enables an

efficient transfer of data traffic as well as fast pro-

tection mechanisms. RPR technology, which was

standardized as IEEE 802.17 RPR, is based on two

symmetric counter-rotating rings that carry data and

control information [12]. Additionally, the ring to-

pology based on RPR is also studied by ITU-T.

Specifically, ITU-T Recommendation X.87 speci-

fiesMultiple Services Ring(MSR) based on RPR

and a way of multi-service provision over RPR[13].

RPR is designed to operate over a variety of physi-

cal layers, including SDH/SONET, Gigabit

Ethernet, DWDM and dark fiber, and is expected

to work over higher-speed physical layers. Some

RPR technology features (distributed control,

scalability in speed and number of nodes, plug-and-

play operation, support various classes of traffic,

advanced protection mechanism, etc.) triggered

many pre-standard installations by some players in

the telecommunications market (e.g., Sprint,

Luminous, Bell Canada, MCI and SUNET). Thefirst major pre-IEEE 802.17 RPR standard deploy-

ments were Dynamic Packet Transport (Cisco Sys-

tems proprietary solution) networks introduced by

Sprint in 1999 and Macedonia Telecom and China

Telecom in 2001.

RPR technology implements the spatial reuse,

which increases the overall aggregate bandwidth

of the ring. Unicast frames are removed from the

ring at their destination, which means that they

occupy bandwidth on the links from source to des-tination only. RPR networks support three class of

traffic. Specifically, IEEE 802.17 RPR supports

three types of services, namely Class A, Class B

and Class C. The Class A service is designed to

support real-time applications that require a guar-

anteed bandwidth and low jitter while the Class B

service is dedicated to near real-time applications

that are less delay-sensitive but that still require

some bandwidth guarantees. Finally, The Class C

service implements the best-effort traffic class. This

service is subject to weighted fairness mechanisms,

which ensure that each station gets its fair share of

the bandwidth available.

Two protection mechanisms may be used: steering

and wrapping, both of which provide fast protection

switching comparable with that of SDH/SONET

networks. Neither of these mechanisms requires dedi-

cated protection resources. RPR protection mecha-

nisms have been designed and optimized to maintain

the network connectivity and to minimize the packet

losses in case of fiber cuts or node failures.

RPR seems to be a promising technology, since

most of the major carriers have actively participated

in the standardization process and have shown much

interest in the evolution of the standard. RPR sys-

tems are seen by many carriers as the inevitable suc-

cessors to SDH/SONET ADM-based rings. Indeed,

RPR network may provide performance-monitoring

features similar to those of SDH and, at the same


6/19



time, maintain Ethernets advantages (e.g., low

equipment cost, high bandwidth granularity and sta-

tistical multiplexing capability). We can note that

inability to operate over multiple rings may impede

implementation of RPR in some areas.

5. Multiprotocol Label Switching

Multiprotocol Label Switching(MPLS) is a connec-

tion oriented packet switching technique providing

mechanisms for engineering network traffic patterns

independently of routing tables. MPLS assigns short,

fixed-length (20-bit) labels to network packets that

describe how to forward them through the network.

In an MPLS environment, the analysis of the packet

header is performed just once, when the packet en-

ters the MPLS domain. Label forwarding tables in

routers store information on where to forward the

packets. Additional information can be assigned with

a label, such as class-of-service (CoS) values that

can be used to prioritize packet forwarding. Usage

of MPLS is not limited to IP networks. It may peer

with ATM or Frame Relay networks. Appropriate

standards were defined by IETF [14], [15], [16]. Label

switched path may be tunneled (extended) in such

networks. This functionality extends capabilities ofIP services. Currently, the two main roles of MPLS

are traffic engineering and Virtual Private Network

support. MPLS provides functional traffic engineer-

ing capabilities required to implement policies that

facilitate efficient and reliable network operations

in an MPLS domain. MPLS decouples the routing

and forwarding functionality. Finding an optimal

routing scenario in presence of constraints imposed

by limited capacity of connections and network to-

pology is facilitated. These capabilities can be used

to optimize the utilization of network resources and

to enhance traff ic oriented performance

characteristics. MPLS TE (MPLS Traffic

Engineering) provides capabilities for traffic

tunneling, load balancing and explicit routing.

Moreover, it eliminates the need for manual setting

up of explicit routes. TE functionality encompasses

also resilience issues. MPLS provides fast protec-

tion and restoration mechanisms. The network re-

covers dynamically from a failure by adapting its

topology to a new set of constraints. MPLS VPNs

do not need a predefined logical or virtual channel

provisioned between two endpoints to establish a

connection between the two endpoints. Traffic of

various users is treated separately within the MPLS

network without the need for encryption or tunnel-

ing at lower layers. MPLS VPNs are scalable (as

opposed to connection oriented Frame Relay (FR)

or ATM VPNs requiring hundreds of virtual chan-

nels for each closed group of users). Moreover,

MPLS provides a capability for consolidation of data,

voice and video services. Each VPN may use its own

independent addressing plan. An incumbent opera-

tor does not need to change its addressing plan whiledeploying an MPLS VPN. MPLS also facilitate

Quality of Service (QoS) assurance but it must be

remembered that putting it on a par with QoS archi-

tectures such as IntServ and DiffServ is a miscon-

ception [17]. Its role is different. IntServ and DiffServ

network models are not dependent on OSI/ISO layer

2 technologies and define a general QoS architec-

ture for IP networks, which can integrate different

transmission technologies in one IP network. MPLS

is another networking technique, like ATM andFrame Relay, defined in layers 2 and 3. MPLS was

originally intended to simplify packet forwarding in

routers rather than to address service quality. Some

features of MPLS can facilitate the QoS assurance.

It can extend IntServ and DiffServ capabilities to a

wider range of platforms beyond the IP environment.

It facilitates offering IP QoS services via FR or ATM

networks. Other MPLS features, such as capabili-

ties for load balancing, flow control, explicit rout-

ing and tunneling are also important from the QoSperspective [17].

6. The architecture of current core and metro-

politan networks

The circuit switched voice traffic was tradition-

ally a major part of the traffic in core and metro-

politan networks. Recently, however, besides the

voice traffic, leased lines service has become an


7/19



important source of profits for networks operators.

Additionally, it usually consumes a large part of

the networks capacity. As the data, which may

be identified with IP traffic, proceeds from

narrowband towards broadband connections, it

starts to play the dominant role in transport

networks. Usually, IP routers are simply connected

by SDH/SONET links with STM-16/OC-48 or

STM-64/OC-192 interfaces. If protection is needed,

the connections are transported over SDH/SONET.

Otherwise, if it is sufficient to provide resilience

purely at the IP layer, the IP router connections are

directly mapped into static WDM wavelength-

based connections. The MPLS technology, which

additionally improves the functionality of the IPlayer, is installed in IP networks today. Resilience

functions are possible at the MPLS layer and may

be implemented on a per service basis in future

networks.Broadband Leased Lines Services are

based on SDH/SONET as well as on WDM

technologies. At the same time, the position of dark

fiber services is continuously decreasing. The trans-

port of voice traffic is mainly performed by SDH/

SONET networks, however, the role of packet

switching technologies, mainly the IP protocol, is

growing. It can be noted that voice traffic may be

conveyed by IP protocol over SDH/SONET tech-

nology or by IP over MPLS over SDH/SONET or

by MPLS without usage of the IP protocol. A way

in which voice payload may be directly encapsu-

lated is defined in [18]. In this paper we use the term

voice over IP to indicate that voice is transported

over packet switching technologies. The protocol

stack for the current transport networks is shown

in Figure 1.

Services

Currently, the core and metropolitan networks pro-

vide Provisioned Bandwidth Service at the IP and

the SDH/SONET layer. Additionally, it is possible

to offer VPN services at the IP, WDM and SDH/

SONET layers. At both, the SDH/SONET and WDM

layers, service provisioning may be very time

consuming, particularly at the WDM layer, where

provisioning of a service is a mostly manual process.

QoS/TE

Referring to the QoS and traffic engineering (TE)

features the situation is nearly the same in the coreand metropolitan networks. At the IP layer, MPLS

supports traffic engineering, but quality of service

parameters are still insufficient for a majority of ser-

vice providers. QoS at the SDH/SONET usually

meets expectations of users.

Connection provisioning

It can be noted that at the IP layer switched connec-

tions may be provisioned while the SDH/SONET

technology allows only permanent and soft-perma-

nent connections. At the WDM layer, permanentmanually configured connections are feasible only.

Hence, connection provisioning may be very

laborious, time consuming and expensive. It seems

that such a functionality, in most cases, is sufficient

for operators of core networks. In contrast, in the

metropolitan networks there is a growing demand

for fast connection provisioning.

Resilience

Protection relies on pre-provisioned backup

resources, whereas restoration, in principle, assigns

backup resources only after the occurrence of a

failure. Currently, both protection and restoration are

possible in the core and metropolitan networks at

the MPLS/IP level, while only protection mecha-

nisms are provided at the SDH/SONET layer.

Drivers

Several factors play a significant role in the evolu-

tion of current transport networks. The growing vol-

ume of data traffic to be transported over networFig.1 Current transport networks - protocol stack


8/19



ks is commonly referred to as an important driver

impacting the transformation of the network

architecture. However, especially in the developed

world, dozens national and international fiber

backbones have been installed. Hence, besides

some developing countries, there is no need for

new fiber links crossing continents. The situation

in the metropolitan areas is similar, even though

it seems that there is still some room for new

installations. Therefore, especially in the short

term perspective, drivers different from those

purely increasing demand for bandwidth are ex-

pected to dominate. Taking into account the world-

wide economic slowdown, which we experienced

in last years, the huge investments done recentlyby telecommunications operators and the strong

competition on the telecommunications services

market, it is obvious that cost reduction will be

the predominant design constraint of the future

transport networks. Spending of telecommunica-

tion operators can be reduced by limiting the nec-

essary capital expenditures (CAPEX) on one side,

and by optimizing the network operational costs

(OPEX) on the other. Possible savings in opera-

tional expenditures together with enhanced net-

work flexibility will be critical for the commer-

cial success of network operators.

SHORT TERM

SCENARIO - ECONFIGURABLE WDM

The first step on the evolution path to cost reduc-

tion and increased network flexibility is the in-

troduction of integrated and reconfigurable WDM

systems (denoted here as rWDM). This step

complements the need for increased bandwidth

and the need for cost reduction at the lowest opti-

cal layer of the transport network. So far, at the

WDM layer there are mostly static cross-connect-

ing elements. It is not possible to allow in-service

selection of the optical channel to be switched,

added or dropped by the use of software control.

Instead of early deployed point-to-point WDM

systems, future systems will deploy a wavelength-

routed network. It may be accomplished by the

use of flexible or reconfigurable optical add/drop

multiplexers, optical cross-connects, as well as

tunable lasers and receivers. Hence, Leased Line

Service and Ethernet will be increasingly basedon the optical layer. Improvement of flexibility

seems to be particularly important in metropoli-

tan networks, whereas in the core, usually

underutilized links do not have to be equipped

with reconfigurable elements.

In the metro environment, some (mainly

newcomer) operators will deploy Gigabit Ethernet

and RPR technology-based networks. Furthermore,

single-wavelength fiber optics systems will play a

minor role in the transport networks. The layeredarchitecture for the short term scenario with

reconfigurable WDM is shown in Figure 2.

Services

At this evolution step, the IP and SDH/SONET

layer service provisioning is still time consuming

and static in both metropolitan and core networks.

However, the introduction of reconfigurable ele-

ments at the WDM layer allows network operators

to offer PBS at the optical level. In contrast to the

current situation, service provisioning at the opti-

cal layer may be performed faster. In addition to

permanent and soft-permanent connections, which

are feasible at the IP and SDH/SONET layers, in

WDM networks with reconfigurable elements it

will be possible to provision connections via the

management plane. Additionally, in the metropoli-

Fig.2 Network Evolution - reconfigurability at

the WDM layer


9/19



tan networks flexibility of services may be im-

proved by the increased number of Ethernet and

RPR installations.

QoS/TE

Traffic engineering aspects do not change at this

stage. Similarly, the QoS remains the same in the

core networks. However, in the metropolitan area

quality of service may be slightly augmented by

implementation of RSTP and RPR.

Connection Provisioning

At this stage, rWDM will boost the connection

provisioning in the core and metropolitan

networks. Additionally, wider, than in the previ-

ous phase, implementation of Ethernet and RPR

will increase the capability to deliver connections/services to customers.

Resilience

At the IP and SDH/SONET layers, resilience re-

mains the same as in the previous scenario, i.e.,

IP uses its rerouting capability in failure cases and

SDH/SONET offers pre-provisioned protection

options. At the reconfigurable WDM layer, at this

stage, it will be possible to perform pre-provi-

sioned protection, i.e., an rWDM device may de-

tect a Loss of Signal (LOS) and automatically

switch traffic from a faulty to a pre-provisioned

working link. However, proper procedures to co-

ordinate protection/restoration mechanisms at the

electrical and the optical layers to provide a sur-

vivable network with QoS support and race con-

ditions avoidance mechanisms have to be

implemented. Similarly as for core networks, re-

silience in some metropolitan networks will be

affected by the introduction of reconfigurability

at the WDM layer. In other networks, the imple-

mentation of RPR may help network operators to

ensure efficient protection at the required level.

Drivers

Strong competition on the market and continuously

decreasing profit margins will force telecommuni-

cation operators to find new customers and to of-

fer new services. This cannot be done using the

business model based on the cost reduction only.

We believe that the offer of new services is the key

for success. The development of applications as-

suring fast and reliable access to remote resources

- data storage applications, network-wide compu-

tation services, virtual reality - will affect network

architectures as well. This can be translated into

technical requirements as a need for flexible and

standard framing methods for a wide range of cli-

ent signals, starting fromFiber ChannelorEnter-

prise Systems Connection (ESCON) formats to

Ethernet or IP protocols. However, at this stage,

there is still a problem with adapting the SDH/

SONET layer for transporting data traffic with ei-

ther block-coded data streams such asFiber Chan-

nelor Fiber Connection (FICON) or packet-ori-

ented data streams, such as IP/PPP or Ethernet.Moreover, the legacy infrastructure in both core and

metropolitan networks does not have the ability to

adjust already established connections to changing

conditions in the network. The equipment and/or

control software are needed to allow a network

operator fast adaptation to needs of a customer. This

is particularly true for metropolitan networks.

Hence, the TDM infrastructure, i.e., the voice ori-

ented technology, has to be adapted to the data cen-

tric environment with proper flexibility and adapt-ability for the changing requirements. The need for

cost effective solutions is still essential for network

operators. The cost reduction may be achieved by

transition of data transport and switching from the

electrical to optical domain. Therefore, the core as

well as metropolitan networks should be continu-

ously transformed towards the optical domain.

MEDIUM TERM SCENARIOS

1. Implementation of Generic Framing Procedure

In this phase, the Generic Framing Procedure (GFP)

will be implemented. The Generic Framing Proce-

dure defines a very effective way of mapping a wide

variety of data signals into transport networks [19]. It

adapts traffic from higher-layer client signals over

SDH/SONET, OTN or dark fiber into a common


10/19



format. The ITU-T recommendation defines two

transport modes. The first mode, referred to as

Frame-Mapped GFP(GFP-F), is optimized for the

adaptation of PDU-oriented streams such as IP, na-

tive PPP, MPLS or Ethernet traffic. The second

mode, optimized for block-code-oriented streams,

is called Transparent GFP(GFP-T). This mode is

used for Gigabit Ethernet, Fiber Channel, FICON

and ESCON traffic. Both transport modes may co-

exist within the same transport channel. GFP ad-

dresses requirements of delay-sensitive applications

such as storage area network (SAN). It is also ex-

pected to support the new IEEE 802.17 RPR

standard. Another advantage of GFP is its particular

suitability to high-speed transmission links stemmingfrom reduction of processing requirements for data

link mappers/demappers as well as simplification of

receiver logic [20]. At this stage, it seems that the cen-

ter of gravity will shift towards services offered

through GFP over WDM rather than SDH/SONET

over WDM. Functionally, GFP consists of common

and client-specific aspects. The former apply to all

traffic. It encompasses data link synchronization and

scrambling, PDU delineation, PDU multiplexing and

client-independent performance monitoring. The

client-specific aspects include mapping of particu-

lar client PDUs into the GFP frame, client-specific

performance monitoring and OA&M functionality.

Interrelation between GFP-F, GFP-T, the client-spe-

cific and common aspects as well as GFP relation-

ship to client signals is shown in Figure 3.

Examples of client payloads that can be mapped

on SDH/SONET via GFP are as follows [21]:

Fiber Channel (850/1062.5 Mb/s);

VC-4-6v/STS-3c-6v (900 Mb/s);

Gigabit Ethernet (1000/1250 Mb/s);

VC-4-7v/ STS-3c-7v (1050 Mb/s); FICON (850/1062.5 Mb/s);

VC-4-6v/ STS-3c-6v (900 Mb/s).

At the same time more efficient use of avail-

able network resources will be achieved. For many

metropolitan areas it seems that there is still some

room for resource usage optimization on the per

day basis. Together with the switching capability,

capacity of links used by business customers dur-

ing the day can be re-used for residential users in

the evening. Such a resource usage optimizationat the medium time scale can be achieved at this

stage by the use of theLink Capacity Adjustment

Scheme (LCAS) protocol with support of agile

management systems. LCAS [22] is an extension

to Virtual Concatenation. It allows the dynamic

alteration of bandwidth of SDH/SONET transport

pipes. This is a key functionality for the transport

of data-traffic coming from IP-applications while

saving bandwidth. The number of concatenated

payloads may be increased or decreased at any

time without affecting traffic currently being sent.

Moreover, LCAS will automatically decrease the

capacity if a member of a VCG experiences a fail-

ure in the network, and LCAS will increase the

capacity when the network recovers. When one

of the constituent channels experiences a failure,

the failed channel will be automatically removed

while the remaining channels are still working.

Thus, the available bandwidth will be lowered but

the connection will be maintained. It can be noted

that such a solution provides a lower probability

of a complete connection failure in the system.

The synchronization between endpoints during the

addition or deletion of channels to a VCG is done

via signaling. Similarly, single-wavelength fiber

optics systems will be less used due to still in-

creasing traffic. Their use will be mostly limited

to access and metro areas. The development of

new telecommunication services will also impose

Fig.3 GFP mapping relationships


11/19



more stringent requirements on methods respon-

sible for providing and controlling services with

guaranteed quality. Presumably, the growing

amount of voice traffic and data traffic with strin-

gent requirements will be conveyed by the IP/

MPLS and Differentiated Services (DiffServ)

networks, which will be introduced at this stage.

Differentiated Services architecture is a solution

for providing different levels of service quality[23]. Independent flows choose one of the limited

number of predefined services. Flows (packets)

that choose the same service are aggregated and

receive the same level of QoS. Aggregated packet

processing by a network node is calledPer Hop

Behavior(PHB). Currently, the DiffServ archi-tecture defines expedited forwarding (EF) [24] and

assured forwarding (AF) PHBs [23] beyond the best-

effort service. Traffic entering a network is clas-

sified and conditioned at the boundaries of the

network. Active queue management mechanisms

within a DiffServ domain are responsible for in-

telligent dropping of packets not conforming to a

contract between a customer and an operator.

In metro environments, it seems that the

Ethernet as well as RPR standard systems will be

widely deployed at this stage. Dynamic develop-

ment of the Ethernet networks will probably es-

sentially impact services offered in packet

switched networks as well. In the medium term

perspective, high speed, widely used and matured

IP networks with MPLS support will be used for

circuit emulation and for transparent transport of

ATM, FR, Ethernet or even SDH data units. Such

a network architecture - the architecture enabling

transfer of layer two data units (e.g., Ethernet) over

layer three (IP or IP/MPLS) may be very interest-

ing for low cost and efficient interconnection of

different network domains in highly competitive

metropolitan environment. The IETF has already

published first RFC standard on architecture of

Pseudo Wire Emulation Edge-to-Edge (PWE3)

services [25]. Next documents are expected to deal

with mapping procedures for encapsulation of

specific technologies, set-up and maintenance of

the tunnel for data encapsulation, traffic policing,

data fragmentation, connection verification and

others. Internet drafts on the enumerated issues

are already available at the web site of PWE3

working group [26]. Taking into account introduc-

tion of PWE3 services, layered architecture of

transport network for the medium term scenario

is shown in Figure 4.

Services

Under this scenario, at both core and metropolitan

networks, at the IP level, Provisioned Bandwidth

Service and Bandwidth on Demand Service may be

offered while in rWDM networks still the former

one only. Due to the introduction of LCAS it is pos-

sible to dynamically increase or decrease the band-

width of a connection at the SDH/SONET layer.

Hence, the SDH/SONET better meets user

requirements. Non-broadband connections such as

STM-1/OC-3, up to now realized by using the SDH/

SONET technology, will be provided by the IP/

MPLS protocol as well. At the optical layer, only

high bandwidth connections may be offered.

Moreover, it seems that at this stage pseudo wire

emulation service will be implemented.

QoS/TE

In the core networks, the QoS remains the same as

in the previous scenario while quality in the metro-

Fig.4 Transport Network Evolution - implemen-

tation of GFP


12/19



politan networks may be enhanced by DiffServ. Traf-

fic engineering, however, may be improved by

implementation of LCAS in both core and metro-

politan networks.

Connection Provisioning

At this stage, the connection provisioning capabil-

ity remains the same as in the previous scenario.

Resilience

In the previous evolution steps, protection and/or

restoration mechanisms were available not only at

the SDH/SONET layer but also realized at the IP

layer using MPLS functions, and at the WDM layer.

In a network based on IP over rWDM with GFP

framing, regardless of type of network, majority of

functions of the next generation SDH/SONET (NG-SDH/SONET) technology, including resilience

aspects, will be distributed over the IP and rWDM

layers. NG-SDH/SONET denotes here SDH/

SONET with the VC and LCAS functionality. The

need for proper coordination of protection/restora-

tion mechanisms is still valid.

2. Introduction of Optical Transport Network

An Optical Transport Network (OTN) is composed

of a set of optical network elements connected byoptical fiber links. An OTN is able to provide func-

tionality of transporting, multiplexing, routing,

management, supervision and survivability of op-

tical channels carrying client signals. A distin-

guishing characteristic of the OTN is its provi-

sion of transport for any digital signal indepen-

dently of client-specific aspects, i.e., it provides

client independence. As such, according to the

general functional modelling described in [27], the

OTN boundary is placed across the Optical Chan-

nel/Client adaptation, in a way to include the

server specific processes and leaving out the cli-

ent specific processes. The client specific pro-

cesses related to Optical Channel/Client adapta-

tion are described in Recommendation G.709 [28],[29]. The standardization process of the OTN is

conducted by the ITU-T. Namely, ITU-T Study

Group 15 has been designated as a Lead Study

Group for two important activities - the project

onAccess Network Transport (ANT) and Optical

Transport Networks & Technologies (OTNT). The

OTNT Standardization Work Plan describes the

activities towards the specification of architectures

and technologies forMetropolitan Optical Net-works (MON), as well asLong Haul Optical Net-

works (LHON) [29]. The main difference between

these two networking domains is the network re-

quirements posed by telecommunications

operators. The main driver forcing the evolution

of metropolitan optical networks is low cost

connectivity. This drives the adaptation of the lo-

cal area network technologies (e.g., Ethernet). On

the other hand, pervasive ring topologies force the

introduction of RPR technology in the metropoli-tan networks. The issue of service dynamics also

has to be considered. An increased demand for

fast provisioned data transmission services char-

acterizes rather metropolitan than long haul opti-

cal networks. The technologies considered to sup-

port MON include SDH/SONET, DWDM/

CWDM, Optical Ethernet, RPR and APON/EPON

(ATM/Ethernet PON) [29]. The most promising

technologies applicable to LHON implementation

include almost the same set of technologies, ex-

cluding probably RPR and APON/EPON. The key

recommendations on the OTN transport plane are

at hand. A framework for OTN as well as refer-

ences for definitions of high-level characteristics

of OTN along with a description of the relevant

ITU-T Recommendations is provided in G.871 [30].

The network architecture is characterized in G.

872 [31]. G.709 defines the interfaces of the opti-

cal transport network to be used within and be-

tween subnetworks of the optical network, par-

ticularly the optical transport hierarchy (OTH),

functionality of the overhead in support of multi-

wavelength optical networks, frame structures, bit

rates and formats for mapping client signals [28].

G.806, G.798 and G. 805 specify the equipment

functionality [32], [33], [27]. At the same time, G.874,

G.874.1 and G.7710 describe equipment manage-

ment functions of transport network elements [34],[35], [36]. Specifications of protection switching in


13/19



OTNs, e.g., G.808.1 and G.873.1, are available

as well [37], [38]. G.8201 and G.8251 are related to

the error performance parameters for multi-opera-

tor international links and the control of jitter and

wander within OTN [39], [40]. The physical issues,

besides [1] and [2], i.e., the physical layerinter-do-

main interface (IrDI) specifications for optical

networks, are provided in G.959.1 [41]. The intro-

duction of the OTN at the optical layer will en-

able telecommunications operators to provide

digital services of controlled quality to the most

important customers, customers requesting high

data rate and high quality services. OTN support-

ing protection at the optical layer is the step at the

network evolution path supporting the demand forhigh quality services, while implementation of

restoration protocols at the same time will addi-

tionally assure better resource usage and prom-

ises cost reduction for offered services. The lay-

ered architecture for the medium term scenario

with OTN is presented in Figure 5.

Services

The implementation of the OTN ensures that digital

optical services may be offered, in contrast to the purely

analogue WDM technology. OTN guarantees client

independence, hence, a wide range of client signals after

GFP encapsulation may be transparently conveyed.

QoS/TE

The introduction of OTN allows network operators

to ensure QoS parameters at the optical layer. This

can be achieved due to the Reed-Salomon 16 byte-

interleaved forward error control(FEC), as de-

scribed in G.709 [28]. Additionally, some proprietary

FEC schemes are allowed and even better param-

eters of optical signal may be achieved. Moreover,

OTN connection monitoring capabilities allow op-

eration in a multicarrier environment. Namely, G.

8201 defines error performance events, parameters

and objectives forOptical Channel Data Unit

(ODUk) paths of the OTN [39].

Soft-permanent and permanent connections will

be provisioned by using the OTN technology.Moreover, at the OTN layer, connection monitoring

will be conducted. Therefore, connection provision-

ing capabilities will be increased. Additionally, due

to introduction of FEC, the strong limitation on some

parameters of optical elements may be reduced.

Hence, more economical network elements may be

used. Furthermore, in contrast to current networks,

longer transparent optical paths may be established.

Resilience

At the optical layer, besides LOS, a link or path deg-radation may be detected and proper mechanisms will

be used to protect data traffic. The process of path

selection for protection/restoration will be presum-

ably performed in the management plane. G873.1

defines the APS protocol and protection switching

operation for the linear protection schemes for the

OTN at the ODUk level [38]. This recommendation

defines subnetwork connection protection with a

sublayer, inherent and non-intrusive monitoring.

Drivers

It seems that at this stage three main drivers will force

the development of both core and metropolitan

networks. Firstly, the voice revenue will dramatically

drop. The growing number of mobile telephony users

on one hand along with the increasing number of cli-

ents using packet voice techniques on the other, will

probably dry up todays main revenue stream. Hence,

the pressure on carriers to find new sources of revenues

will grow. Secondly, the number of broadband users

Fig.5 Transport Network Evolution - OTN imple-

mentation


14/19



will likely multiply. Ubiquitous broadband service of-

fered to thousands of clients may force operators to

transform their infrastructures towards networks with

data centric technologies only. Additionally, broadband

services may cannibalize the traditional voice service

offered by operators by common usage of packet

telephony, which may hasten the withdrawal from fixed

telephony. Thirdly, it seems that the market for corpo-

rate telecom service will grow. Hence, network op-

erators may start to offer new and more intelligent

services. They can take part in the growing trend to-

wards outsourcing and offer, for example, not only

dumb connections but a whole package of services. It

can be hosting the IT equipment, the management of

data centers, the backup or disaster recovery serviceor a new comprehensive service. Such a single inte-

grated service may be combination of knowledge

about networks, possessed by network operators, and

skills related to software integration, which is per-

formed today by the IT sector. An example of such a

convergence is the voice over IP (VoIP) technology.

Undoubtedly, the key issue is the ability to offer flex-

ible and customer tailored services. Hence, automated,

or so-called intelligent network is a matter of ut-

most importance.

LONG TERM SCENARIOS

1. Implementation of Automatically Switched

Optical Network

The introduction of intelligence (by means of sig-

naling and routing protocols) in multilayer optical net-

works will enable network operators to meet emerging

requirements, such as: dynamic and rapid provision-

ing of connections, automatic topology discovery and

network inventory, reactive traffic engineering, and

faster optical restoration. All these functions and fea-

tures are important for the implementation of cost

optimized, high quality telecommunication services

offered in a flexible, high data rate telecommunication

network. TheAutomatically Switched Optical Network

(ASON), and its more generic counterpart, i.e.,Auto-

matic Switched Transport Networks (ASTN), are a set

of control plane components which provide the possi-

bility of setting up, maintaining and releasing connec-

tions [42], [43]. By using ASON, networks operators will

be able to offer services which may be initiated by a

client through the UNI interface [42]. ASON as well as

ASTN, which are being developed by Study Group 15

of ITU-T, is the architecture that defines components

and a set of reference points and rules which must be

applied at the interface between clients and the net-

work as well as between networks. The architecture

defined by ITU-T is protocol independent and suffi-

ciently generic to support various business

requirements. The control model assumed in the archi-

tecture is the overlay model while connections may bysignaled or may be provisioned in a hybrid way [42].

ITU-T Recommendation G.7714 describes the speci-

fications for automatic discovery techniques to aid re-

source management and routing in the ASON networks[44]. G.7713 provides the requirements for the distrib-

uted call and connection management for both the User

Network Interface (UNI) and theNetwork Node Inter-

face (NNI) [45]. G.7713.1 is the answer for the require-

ments provided in [45] and is based on the PNNI/Q.2931[46]. Meanwhile, G.7713.2 meets the same requirements

but is based on the RSVP-TE [47]. In G.7715, the re-

quirements and architecture for the ASON routing func-

tions used for the establishment ofswitched connec-

tions (SC) andsoft-permanent connections (SPC) are

specified [48]. However, the protocol-neutral require-

ments for a hierarchical link state routing protocol are

provided in a newly proposed G.7715.1 [49]. The trans-

port of distributed call and connection management and

signaling messages may be performed by a data com-

munication network (DCN) described in [50]. The Opti-

cal Internetworking Forum proposed the User Network

Interface (UNI) 1.0 Signaling Specification [51]. OIF is

a non-profit organization with the aim to foster devel-

opment and deployment of interoperable products and

services for data switching and routing using optical

networking technologies. The organization, which has

the official liaisons with ATM Forum, IEEE 802.3,

IETF and ITU-T SG 15, has six working groups. The

groups cover a wide range of technical issues related


15/19



to optical networks. OIF Implementation Agreement

OIF-CDR-01.0 specifies the usage of measurement

functions that an Optical Switching System will need

to perform in order to enable carriers to bill for OIF

UNI 1.0 optical connections using their legacy billing

systems [52]. It also specifies three formats for storing

these usage records in files for processing by the

carriers billing systems. OIF-SEP-01.1 defines a com-

mon Security Extension for securing the protocols used

in UNI and NNI [53]. The OIF-E-NNI-01.0 specifies of

External NNI(E-NNI) signaling abstract messages,

attributes, and flows for end-to-end dynamic establish-

ment of transport connections across multiple control

domains and, so far, applies to SDH/SONET connec-

tion services only[54]

. It can be noted that with ASON,Generalized MPLS(GMPLS) family protocols may

be used as well, e.g.,Resource Reservation Protocol -

Traffic Engineering(RSVP-TE) [55]. On the other hand,

there are some differences which can make the pro-

cess of reusing GMPLS tools in the ASON scenario

difficult for networks operators. Firstly, UNI is not a

trusted reference point, and hides all routing and ad-

dressing information pertaining to the interior of the

network from the user. Moreover, a user belongs to a

different address space than the internal network nodes.

Hence, the ASON scenario may be identified with the

overlay model only. Secondly, the ASON concept as-

sumes a distinction between call and connection sig-

naling which is not present in the GMPLS set.

Therefore, from todays point of view, in some areas

the GMPLS set is well suited to operate over the ASON

architecture and at the same time some mechanisms

taken from the ITU-T and IETF standardization seem

incoherent. Furthermore, ASON focuses merely on

SDH/SONET, OTN and PDH while GMPLS embraces

packet, time-division, wavelength and spatial switching.

However, the authors believe that a solution based on

a constructive compromise between ITU, IETF as well

as OIF will be found.

Essential advances in optical technology will prob-

ably enable a new transfer mode of data in optical

networks. In the long term scenario, at the edge of next

generation optical networks, data addressed to a par-

ticular destination will be collected together and formed

into an optical data unit, referred to as burst. The burst

is to be forwarded towards destination with the use of

any available wavelength. All the control information

necessary to transfer the burst to the final destination

will be sent with by using an out of band control chan-nel (next wavelength, for example). The optical net-

work adapting the presented idea is referred as Optical

Burst Switching(OBS) network. Bursts, composed of

data of distinct users, can share in an OBS network a

single wavelength. It is expected that OBS networks

will offer much higher flexibility, will increase network

resource utilization ratio and will essentially improve

efficiency of the optical and IP network interface. In

the perspective of long term scenario it is expected that

some open issues specific to OBS networks will besolved, for example, optical routing protocols, QoS

assurance, burst assembly procedures, resource

reservation, as well as security issues. The layered ar-

chitecture of future transport networks with possible

implementation of the OBS idea in the transport plane

is shown in Figure 6.

Services

The implementation of ASON or GMPLS in both core

and metropolitan networks significantly changes the

spectrum of services offered to customers. Under this

scenario,Bandwidth on Demand Service (BDS) is in-

troduced at the optical layer. Therefore, a client may

Fig.6 Transport Network Evolution - implementation of

ASON or GMPLS


16/19



request a connection through the UNI interface.

However, under UNI 1.0 specification it is possible to

create and delete SDH and SONET connections only.

The authors believe that a later version of the UNI speci-

fication will improve the functionality of the current

UNI interface. Additionally, ASON enabled transport

networks will allow network operators to offer optical

VPNs created and modified on demand.

QoS/TE

At this stage, the QoS issues remain the same as in

the previous scenario. However, the implementation

of ASON or GMPLS in the control plane improves

significantly traffic engineering in optical networks.

Namely, the infrastructure of core and metropolitan

networks may be effectively and dynamicallyadapted to changing conditions. Though, it seems to

be too early to specify how it will be performed.

Connection provisioning

A switched optical network will provide bandwidth and

connectivity to an IP network in a dynamic manner (i.

e., based on current demand patterns) compared to rela-

tively static services available at the previous evolu-

tion steps. Under this scenario, all types of connections,

i.e., permanent, soft-permanent and switched connec-

tions may be provisioned at the optical layer. However,it is assumed that a connection request at UNI orEx-

ternal Network to Network Interface (E-NNI) will con-

tain only the requested Class of Service and not the

explicit protection and restoration type [42]. At this stage,

the overlay model may be used only. Hence, separate

protocols and/or separate instances of the same proto-

col exist in the control plane for each layer. It can be

noted that at the IP and optical level, routing and sig-

naling mechanisms used here will be derived from the

GMPLS set. An interesting option for the ASON ar-

chitecture is the usage of thePrivate Network to Net-

work Interface (PNNI) protocol instead of the GMPLS

set. PNNI has an advantage over GMPLS in the sense

that it is a mature solution. Moreover, its characteristics,

e.g. support for CoS and QoS, protection and

restoration,Permanent Virtual Connection (PVC), Soft

Permanent Virtual Connection (SPVC) and Switched

Virtual Connection (SVC), make PNNI well suited for

the ASON concept. Additionally, contrary to the

GMPLS protocol family, PNNI provisions Call Ad-

mission Control(CAC) functions. On the other hand,

there are some drawbacks which may prevent network

operators from using PNNI. Firstly, PNNI have to be

adjusted to operate over the ASON architecture.

Secondly, PNNI signaling messages are exchanged in-

band only. Thirdly, it usesNetwork Access Service Point

(NSAP) addresses and not ubiquitous IP addresses.

Moreover, it is not as common as the Internet Protocol.

Therefore, special mechanisms have to be standard-

ized and internetworking devices installed to translate,

e.g., RSVP messages into PNNI. Additionally, the us-

age of PNNI is inconsistent with a general trend to-

wards convergence of IP and optical networks.

However, so far the question which solution, PNNI orthe GMPLS set, will be used is still open.

Resilience

So far, protection and restoration mechanisms in

optical networks with dynamically provisioned con-

nections are not specified. However, intensive stan-

dardization processes carried out by ITU-T, and other

fora, indicate that specifications related to resilience

in ASON networks will be provided soon.

2. Implementation of Generalized MultiprotocolLabel Switching

Generalized Multiprotocol Label Switching(GMPLS)

is a tool by which various electrical and optical

elements, e.g., routers, switches, add-drop multiplex-

ers or cross-connects may be commonly controlled.

GMPLS extends MPLS to encompass time-division,

wavelength and spatial switching (e.g., incoming port

or fiber to outgoing port or fiber) [56]. Hence, by the

usage of GMPLS, the signaling and routing part of

the control plane will be facilitated in comparison to

the previous scenario, where independent control

plane tools for each layer exist (see Fig. 6). Moreover,

it should also reduce operating costs. GMPLS extends

intra-domain link-state routing protocols already ex-

tended for TE purposes, i.e., OSPF-TE and IS-IS-TE

as well as proposes a suitable signaling protocol, i.e.,

RSVP-TE [55]. The GMPLS scenario differs from

the previous one in that, here, a client has visibility of


17/19



topology of a provider and may participate in the pro-

cess of setting up a connection. Hence, connection

set-up may also be the responsibility of the end user.

It can be noted that under this scenario many mecha-

nisms remain the same as in the previous case.

Namely, at the electrical layer the same protocols may

be used and some changes are necessary at the opti-

cal layer only, i.e., signaling and routing at the UNI

and NNI interface have to be improved. Therefore,

implementation of GMPLS in the peer model in-

volves a more significant change of policy for net-

work operators rather than changes of the

technology. Certainly, this scenario may be used by

some operators earlier, e.g., newcomers, while for

others it may be unacceptable, especially for a shortterm perspective. On the other hand, proper proce-

dures for automatic connection negotiation at the

user-to-network as well as the network-to-network

interface have to be standardized. It can be noted

that both IETF and ITU-T are very active which

holds promise that proper standards will be avail-

able soon. By using GMPLS signaling it will be pos-

sible to offer PBS, BDS as well as VPNs at various

levels (packet, TDM, wavelength and fiber). In

addition, a user may participate in the process of

setting up connections, although it imposes higher

requirements on users equipment. Referring to the

connection provisioning, the same functionality will

be available as in the previous case. The usage of

GMPLS allows to achieve smooth coordination be-

tween protection and restoration mechanisms in both

electrical and optical layer, because a single control

instance is aware of the status and the resources of

all network layers. The protection and restoration

level may be chosen in an optimized way. However,

the coordination of the resilience functions on all

layers involved is a complex task and needs a fur-

ther study. Moreover, looking at another dimension

of these mechanisms, so far, MPLS protection and

restoration is being standardized for intra-domain pur-

poses only. It can be noted that with the GMPLS pro-

tocol family it is possible to use the overlay, the aug-

mented as well as the peer model. Although the con-

trol plane evolution scenario may vary depending on

type of network operator (e.g., incumbent, newcomer),

it is quite possible that the control plane will evolve

from overlay, through augmented, to the peer model.

CONCLUSION

In the paper, we addressed the problem of construct-

ing an evolution path for transport networks. At the

beginning, we presented main networking technolo-

gies and the current status of transport networks.

Additionally, we provided the readers with the main

limitations of todays networks. Motivated by the

need to enhance both metropolitan and core networks

we outlined the evolution path towards future

solutions. The path was provided in three stages,

namely, short, medium and long term. We believe

that it is potentially possible, under some conditions,

that these stages may be identified with one-, three-,

and five-year time scale, respectively. The main fo-

cus was on the data and control planes. This evolu-

tion path was referred to the standardization pro-

cesses in the main standardization bodies (e.g., ITU-

T and IETF). We evolved our proposals in two

directions. On one hand, we indicate that substan-

tial progress towards transport networks with im-proved functionality by the enhancement of already

installed equipment may be achieved. On the other

hand, we propose the introduction of new

technologies, e.g., OTN or ASON, in later evolu-

tion steps for further optimization and transforma-

tion of core and metropolitan networks. We discussed

the proposed evolution steps regarding technical

aspects like offered services and resilience functions.

With the ever increasing amount of data traffic it is

seen as necessary to provide higher switching

granularities by the optical layer and increased flex-

ibility in the transport networks on both the range of

acceptable client formats as well as configuration dy-

namics for fast provisioning of new services and effi-

cient networking functions such as resilience. Depend-

ing on the special requirements of the network pro-

vider and the business case, different concepts have

been identified. The implementation of GMPLS based

networks promises to be most cost efficient but is not


18/19



suitable to all network scenarios.

ACKNOWLEDGEMENTS

The authors wish to thank all participants of IST

Project LION and the European Commission that

was partially funding the project.

This work was supported in part by the Polish

Ministry of Science and Information Society Tech-

nologies under Grant No. 4 T11D 012 25.

REFERENCES

[1] Spectral grids for WDM applications: CWDM wave-length grid, ITU-T Recommendation G.694.2, Dec. 2003

[2] Spectral grids for WDM applications: DWDM

frequency grid, ITU-T Recommendation G.694.1,

June 2002

[3] Network node interface for the synchronous

digital hierarchy (SDH), ITU-T Recommendation G.

707/Y.1322, Dec. 2003

[4] Architecture of transport networks based on

the synchronous digital hierarchy (SDH), ITU-T

Recommendation G.803, Mar. 2000

[5] Optical interfaces for single-channel STM-64,

STM-256 and other SDH systems with optical

amplifiers, ITU-T Recommendation G.691, Dec. 2003

[6] Cavendish D., Murakami K., et al.: New

Transport Services for Next-Generation SONET/

SDH Systems, IEEE Communications Magazine,

May 2002, Vol. 40, No. 5

[7] Carrier sense multiple access with collision de-

tection (CSMA/CD) access method and physical layer

specifications, IEEE Standard 802.3-2002 (Revision

of IEEE Std 802.3, 2000 Edition), Mar. 2002

[8] Carrier sense multiple access with collision

detection (CSMA/CD) access method and physical

layer specifications, Amendment: Media Access

Control (MAC) Parameters, Physical Layers, and

Management Parameters for 10 Gb/s Operation,

IEEE Std 802.3ae-2002 (Amendment to IEEE Std

802.3-2002), Aug. 2002

[9] The IEEE 802 LAN/MAN Standards Commit-

tee web site, Available: http://www.ieee802.org

[10] Media Access Control (MAC) Bridges, IEEE

Std. 802.1D-2004, (Revision of IEEE Std. 802.1D-

1998), Jun. 2004

[11] Ethernet Services Model, Phase 1, TechnicalSpecification Metro Ethernet Forum 1.0, Nov. 2003

[12] Resilient packet ring (RPR) access method

and physical layer specifications, IEEE Std 802.17-

2004, Sept. 2004

[13] Multiple Services Ring Based on RPR, ITU-

T Recommendation X.87/Y.1324, Oct. 2003

[14] E. Rosen et al., Multiprotocol Label Switch-

ing Architecture, IETF RFC 3031, Jan. 2001

[15] E. Rosen et al., MPLS Label Stack Encoding,

IETF RFC 3032, Jan. 2001[16] D. Awduche et al., RSVP-TE: Extensions to

RSVP for LSP Tunnels, IETF RFC 3209, Dec. 2001

[17] Gozdecki J, Jajszczyk A, Stankiewicz R.:

Quality of service terminology in IP networks, IEEE

Comm. Mag. 41 (3): 153-159, Mar 2003

[18] Voice over MPLS - Bearer Transport Imple-

mentation Agreement, MPLS Forum 1.0, Jul. 2001

[19] Generic Framing Procedure, ITU-T Recom-

mendation G.7041/Y.1303, Dec 2003

[20] E. Hernandez-Valencia, M. Scholten, Z. Zhu,

The Generic Framing Procedure (GFP): An

Overview, IEEE Communications Magazine, May

2002, Vol. 40, No. 5

[21] Bonenfant P. Rodriguez-Moral A.: Generic

framing Procedure (GFP): The catalyst for Efficient

Data over Transport, IEEE Communications

Magazine, May 2002, Vol. 40, No. 5

[22] Link Capacity Adjustment Scheme for Vir-

tual Concatenated Signals, ITU-T Recommendation

G.7042/Y.1305, Feb. 2004

[23] S. Blake et al., An Architecture for Differ-

entiated Services, IETF RFC 2475, Dec. 1998

[24] Davie B. et al., An Expedited Forwarding

PHB, IETF RFC 3246, Mar. 2002

[25] S. Bryant, P. Pate, Pseudo Wire Emulation

Edge-to-Edge (PWE3) Architecture, IETF RFC

3985, Mar. 2005

[26] IETF Pseudo Wire Emulation Edge to Edge

(pwe3) Working Group, Available: http://www.ietf.


19/19


org/html.charters/pwe3-charter.html

[27] Generic functional architecture of transport

networks, ITU-T Recommendation G.805, Mar. 2000

[28] Interfaces for the optical transport network (OTN),

ITU-T Recommendation G.709/Y.1331, Mar. 2003

[29] ITU-T Study Group 15 - Optical and other

transport networks web site, Available: http://www.

itu.int/ITU-T/studygroups/com15/index.asp

[30] Framework for optical transport network

Recommendations, ITU-T Recommendation G.871/

Y.1301, Oct. 2000

[31] Architecture of optical transport networks,

ITU-T Recommendation G.872, Nov. 2001

[32] Characteristics of transport equipment - De-

scription methodology and generic functionality,ITU-T Recommendation G.806, Feb. 2004

[33] [31] Characteristics of optical transport net-

work hierarchy equipment functional blocks, ITU-

T Recommendation G.798, Jun. 2004

[34] Management aspects of the optical transport

network element, ITU-T Recommendation G.874,

Nov. 2001

[35] Optical transport network (OTN): Protocol-

neutral management information model for the net-

work element view, ITU-T Recommendation G.874.

1, Jan. 2002

[36] Common equipment management function

requirements, ITU-T Recommendation G.7710/Y.

1701, Nov. 2001

[37] Generic protection switching - Linear trail

and subnetwork protection, ITU-T Recommendation

G.808.1, Dec. 2003

[38] Optical Transport Network (OTN): Linear

protection, ITU-T Recommendation G.873.1, Mar. 2003

[39] Error performance parameters and objectives

for multi-operator international paths within the

Optical Transport Network (OTN), ITU-T Recom-

mendation G.8201, Sep. 2003

[40] The control of jitter and wander within the

optical transport network (OTN), ITU-T Recommen-

dation G.8251, Nov. 2001

[41] Optical transport network physical layer

interfaces, ITU-T Recommendation G.959.1, Dec. 2003

[42] Architecture for the automatically switched

optical network (ASON), ITU-T Recommendation

G.8080/Y.1304, Nov. 2001

[43] Requirements for automatic switched trans-

port networks (ASTN), ITU-T Recommendation G.

807/Y.1302, Jul. 2001

[44] Generalized automatic discovery techniques,

ITU-T Recommendation G.7714/Y.1705, Nov. 2001

[45] Distributed call and connection management

(DCM), ITU-T Recommendation G.7713/Y.1704,

Dec. 2001

[46] Distributed call and connection management

(DCM) based on PNNI, ITU-T Recommendation G.

7713.1/Y.1704.1, Mar. 2003

[47] Dis tr ibuted Cal l and Connect ion

Management: Signalling mechanism using GMPLSRSVP-TE, ITU-T Recommendation G.7713.2/Y.

1704.2, Mar. 2003

[48] Architecture and requirements for routing in

the automatically switched optical networks, ITU-T

Recommendation G.7715/Y.1706, Jun. 2002

[49] ASON Routing Architecture and requirements

for Link State Protocols, ITU-T Recommendation

G.7715.1/Y.1706.1, Feb. 2004

[50] Architecture and specification of data com-

munication network, ITU-T Recommendation G.

7712/Y.1703, Mar. 2003

[51] Optical Internetworking Forum (OIF) User

Network Interface (UNI) Specification 1.0 Release

2, Feb. 2004

[52] Call Detail Records for UNI 1.0 Billing, Op-

tical Internetworking Forum Implementation Agree-

ment OIF-CDR-01.0, Apr. 2002

[53] Security Extension for UNI and NNI, Opti-

cal Internetworking Forum Implementation Agree-

ment SEP-01.1, May 2003

[54] Intra-Carrier E-NNI Signaling Specification

Implementation Agreement OIF-E-NNI-01.0, Feb. 2004

[55] L. Berger, Generalized Multi-Protocol La-

bel Switching (GMPLS) Signaling Resource

ReserVation Protocol-Traffic Engineering (RSVP-

TE) Extensions, RFC 3473, Jan. 2003

[56] E. Mannie, Generalized Multi-Protocol La-

bel Switching (GMPLS) Architecture. IETF RFC

3945, Oct. 2004

Documents

SDH Layered Architecture