Professional Documents
Culture Documents
Study Paper On Evolution of High Capacity Optical Network Final
Study Paper On Evolution of High Capacity Optical Network Final
on
Evolution of High Capacity Optical
Transport Network
2
Abstract
Presently network traffic is packet-based, generated by a multitude of services and
applications in bursty, unpredictable traffic patterns with widely varying demands on
bandwidth and data transmission performance. Transport technologies, such as
Synchronous Optical Networking (SONET) and Synchronous Digital Hierarchy (SDH),
were not designed for packet-dominated, high-capacity services requiring transmission
capacities of 40 Gb/s and above. With increasing capacity, B100G coherent transport
represented by single-carrier 200G/400G was need of the hour.
In view of above, this paper deals with the existing Optical Transport Network (OTN)
technology and its status B100 G, the challenges faced and application for B100G.
1.0 Introduction
3
Telecommunications industry and service provider networks are quickly evolving to
deal with an explosion of digital traffic driven by multimedia services, mobile
applications, social media, Voice-over-Internet Protocol (VoIP), and cloud
computing, and an ever-growing array of bandwidth hungry applications. For
decades, service provider network traffic was dominated by voice calls, in which
traffic was carried over circuit based networks in a predictable network connection
between pairs of endpoints. Most network traffic today is packet-based, generated
by a multitude of services and applications in bursty, unpredictable traffic patterns
with widely varying demands on bandwidth and data transmission performance.
Service provider networks that were once optimized for voice traffic are now in need
of a new transport technology that can handle modern network traffic patterns and
content.
Wireless services are one of major business sectors of operators. Compared with the
4G era, it is generally agreed in the industry that the bandwidth demand in the 5G
era will increase 100 to 1000 fold. Considering the unified transport of wireline and
wireless networks in the 5G era, the MAN convergence layer will require multiple 100
Gbps channels, and the MAN core layer will even require a bandwidth of more than
10 Tbps. It can be said that 5G development will bring another round of explosive
bandwidth growth. It is a general trend to deploy B100G OTN at both the MAN core
and backbone layers.
The table 1 deals with evolution of various Optical Transmission technologies with
their prominent time period and limitations.
4
required to demultiplex down to the level
where your tributary channel lies. This was
considered as the major in-efficiency.
Lack of OAM: Because of no agreed standard
for management or to monitor the
performance.
Lack of a standardized definition: There was
no standard for PDH rates greater than 140
megabits per second. An alternative was
needed.
SDH Late 80s Upto 40 SONET/SDH cannot scale beyond 40 Gbps
(Synchronous and early Gbps and has problems handling high-bandwidth
Digital 90s clients such as video.
Hierarchy) SONET/SDH has limited Forward Error
Correction (FEC) capabilities. FEC is an error
correction method required to transmit 10
Gbps signals or greater over any reasonable
distance.
SONET/SDH standards could not be extended
into WDM networks, including: Standards for
mapping client traffic on to wavelengths, such
as video, storage, and GDPS mapping
standards. The absence of client mapping
standards meant that signals had to be de-
multiplexed into their basic elements between
the boundaries of carriers using incompatible
equipment.
5
Figure 1: OTN Evolution
6
communications channels, latency measuring, and a standard mapping structure for
multiplexing low-rate signals onto high-speed payloads.
Although it’s now common to link OTN and Ethernet technologies, OTN was not
originally created to work specifically with Ethernet. In fact, OTN was developed to
manage Wavelength Division Multiplexing (WDM) wavelengths with SONET/SDH as
the client payload, given the wide deployment of SONET/SDH at the time. OTN was
also intended to support a manageable wholesaled wavelength infrastructure. It is
this original use case from which the capability of full payload transparency
originated. By 2009, it was clear that the majority of traffic carried by OTN would be
Ethernet-based, so OTN standards were enhanced to closely align with Ethernet
traffic characteristics.
In the 2009 update, G.709 was enhanced to more tightly integrate with Ethernet
data rates and packet formats. As a result, OTN and Ethernet are now inseparable
in most networks. This symbiotic relationship makes OTN the ideal protocol for
transport of Ethernet over Dense Wavelength Division Multiplexing (DWDM)
networks.
7
Figure 2: Optical Transport Module
The Optical channel Payload Unit (OPU) contains the payload frames. The ‘service
layer’ represents the end-user services such as GbE, SONET, SDH, FC, or any other
protocol. For transparently mapped services such as ESCON, GbE, or FC, the service
is passed through a Generic Framing Procedure (GFP) mapper.
The Optical channel Data Unit (ODUk, where k = 1/2/2e/3/3e2/4) contains the OPU
plus overhead. The Optical Transport Unit (OTUk, where k = 1/2/2e/3/3e2/4) contains
the ODU, provides the section-level overhead such as BIP8, and supports the General
Communication Channel (GCC) bytes for overhead communication between network
nodes. The GCC is used for OAM functions such as performance monitoring, fault
detection, and signaling and maintenance. The physical layer maps the OTU into a
wavelength and the Optical Channel (OCh), which runs across the optical line.
An Optical Multiplex Section (OMS) sits between two devices and can multiplex
wavelengths onto a fiber, as shown in Figure 3. An Optical Transmission Section (OTS)
consists of the fiber between anything that performs an optical function on the signal.
An Erbium-Doped Fiber Amplifier (EDFA) counts as ‘line amplifying’ equipment.
8
OTN offers six levels of tandem connection monitoring that enable a network
operator to monitor a signal as it passes through other operators’ networks. This
functional breakdown aids in fault management, as OTN overhead is rigorously
aligned with these points.
9
Figure 4 illustrates how different services are mapped onto common wavelengths
(an OCh always contains a single OTU), thereby providing for sub-wavelength
bandwidth management and decoupling of service rates from the line rate.
The OPUk (k=0,1,2,2e,3,4,flex) are the same information structures, but with
different client signals. The OPUCn has a different information structure than the
OPUk; the OPUCn information structure consists of n times the information
structure of the OPU while the OPUk contain a single instance of the OPU
information structure.
10
Figure 5: OTN Mapping Hierarchy
11
6.0 Advantage of OTN
Primary advantages of OTN include:
Reduction in transport costs: By allowing multiple clients to be transported on a
single wavelength, OTN provides an economical mechanism to fill optical network
wavelengths.
Efficient use of optical spectrum: OTN facilitates efficient use of DWDM capacity by
ensuring fill rates are maintained across a network using OTN switches at fiber
junctions.
Determinism: OTN dedicates specific and configurable bandwidth to each service,
group of services, or each network partition. This means that network capacity and
managed performance (throughput, latency, jitter, and availability) are guaranteed
for each client, and there is no contention between concurrent services or users.
Virtualize network operations: The ability to partition an OTN-switched network into
private network partitions, also referred to as Optical Virtual Private Networks (O-
VPNs), provides a dedicated set of network resources to a client, independent of the
rest of the network. Each network tenant sees only the resources associated with
that tenant’s private partition. Other resources associated with other tenants will
not be visible. O-VPNs also ease network evolution because network upgrades can
be tested or introduced in a protected network partition or ‘sandbox,’ without
the risk of impacting day-to-day network operations in production partitions.
Flexibility: OTN networks give operators the ability to employ the technologies
needed now to support transport demands while enabling operators to adopt new
technologies as business requirements dictate.
Secure by design: OTN networks ensure a high level of privacy and security through
hard partitioning of traffic onto dedicated circuits. This segregation of network traffic
makes it difficult to intercept data transferred between nodes over OTN-channelized
links. And because OTN-switched networks keep all applications and tenants
separate, organizations can effectively stop hackers who access one part of the
network from gaining access to other parts of the network.
Robust yet simple operations: OTN network management data is carried on a
separate channel, completely isolated from user application data. This means OTN
12
network settings are much more difficult to access and modify by gaining
admittance through a client interface port.
7.0 ODUflex
Until the third version of G.709, released in December 2009 by the ITU, only a
handful of ODU rates had been defined to support the main non-OTN client
signals like STM-16/64/256 and 1/10/100 Gigabit Ethernet and lower rate ODU to
higher rate ODU multiplexing. Transport of many additional non-OTN clients
such as Fibre Channel and Video signals as well as variable rate packet fl ows had
also been examined to ensure the continuing effectiveness of the OTN in carrier
networks. The existing ODU rates were not efficient to transport these new
clients, but it was also not desirable to defi ne a new, fixed rate ODU type for
each one. Consequently, the concept of a flexible rate ODU, or ODUflex, was
devised to fill in the gaps of the fixed rate hierarchy and included in the third
version of G.709.
13
all available payload bytes of the ODU carry client data. From a client clock jitter
generation and suppression perspective, a synchronous mapping is preferred.
In this way, mapping of GFP frames into an ODUflex is exactly the same as
mapping GFP frames into a fixed rate ODUk (k = 0,1,2,2e,3,4). For a given ODUfl
ex (GFP), the rate is established and the packet flow is rate adapted into it using
GFP. It is also possible to adjust the rate of an ODUflex (GFP) to address changes
in traffic profiles. Before, during and after this adjustment process, GFP always
14
takes care of adapting the packet flow rate into the available payload bandwidth
of the ODUflex (GFP).
With the rapid development of 5G, big video, and high-performance private line
services, as well as the growing demands for DCI interconnection brought by cloud
networks, demand for higher optical network bandwidth is witnessing exponential
growth. Accordingly, B100G coherent transport represented by single-carrier
200G/400G has become a hot topic.
15
course of developing the B100G standard. These challenges and objectives
included:
• The Shannon channel capacity limits are catching up to optical transport network
capabilities. Specifically, the 50 GHz channel spacing of the standard wavelength
grid currently used for dense wavelength division multiplexing (DWDM) imposes
limits on transporting signals over reasonable distances when they have rates much
over 100Gbit/s. (200Gbit/s was the practical limit per wavelength for distances of
interest in telecom networks at beginning of the B100G project, although higher
rates per wavelength will become possible driven by advances in Coherent DSP
technology and new higher order modulation formats.)
• The old paradigm of adding new discrete rates for OTN had largely reached its
practical limits, making a modular rate and frame structure approach more
attractive.
• The IEEE 802.3bs Task Force working on 400Gbit/s Ethernet was examining several
new approaches that were different from its previous interfaces. The new OTN
format needed to not only carry 400GbE, but also re-use its technology and PHY
components in order to benefit from the Ethernet component cost curves.
• The higher bit rates and increased use of multi-lane interfaces poses additional
considerations regarding Forward Error Correction (FEC) and performance
monitoring.
• Continuing to use 1.25Gbit/s Tributary Slot (TS) sizes in OTN would be impractical
for B100G rates, so a larger Tributary Slot size was desirable.
• The high data rates and the introduction of new data client signals such as the
OIF’s Flexible Ethernet (FlexE) have made the current byte-oriented mapping
16
approach for data clients impractical. This motivated the desire for a wide-word
type of mapping.
• In order to optimize the use of each wavelength, including the transmission reach,
there was a desire to transmit the OTN signals at the rate required for the client
payload being carried rather than at the full discrete rate of the OTUk signal.
A key early decision was that the new B100G signals would not be switched within
the network, since the ability to switch would make them a new layer network for
the carrier to manage. Instead, B100G was defined for only point-to-point
connectivity, with the client signals carried within the B100G signal being switched.
Another closely related foundational agreement is related to the need to be able to
use multiple wavelengths when transmitting an OTUCn signal. It was agreed that
the entire group of wavelengths associated with an OTUCn interface signal would
go through the same fiber and optical switches (i.e., the same Optical Multiplex
Section trails) such that the OTUCn signal can be managed as a single entity.
Consequently, very limited deskew is required if multiple wavelengths are used.
Note that with the recently revised ITU-T nomenclature, a WDM wavelength used
to carry an OTN signal is referred to as an Optical Tributary Signal (OTSi). An Optical
Tributary Signal Group (OTSiG) is the set of OTSi signals that support carrying a
single client such as an OTUCn.
17
The OTUCn, ODUCn, and OPUCn overhead fields are essentially the same as for
the previous generation of OTN. Overhead that pertains to the entire interface (e.g.,
Trail Trace Identifier for proper connectivity checking, remote defect indications,
and delay measurement overhead) only appear on the first OTUC/ODUC of the
OTUCn/ODUCn rather than on all n slices.
The TS size was chosen to be nominally 5 Gbit/s, which gives (100n/5) = 20n TS in
an OPUCn. This TS rate provided the desired coarser granularity, and was the lowest
rate that would efficiently accommodate the new data client signals, including 25
GbE and 16 Gbit/s Fiber Channel. Newer data clients typically have rates that are
divisible by 5 Gbit/s. Legacy data client signals with rates below 5 Gbit/s will be first
mapped into a legacy OTN signal. The exact TS and OPUCn payload rates are
shown in Table 3.
18
Figure 6: Optical Transport Module for B100G
The OPUflex rate is a constant rate that is set to be higher than the maximum
client rate. Rate adaptation between the packet client data and the OPUflex is
performed by inserting Ethernet Idle code words into the 64B/66B stream as
19
defined in IEEE 802.3 for Ethernet. For that reason, the new approach was given
the name “ODUflex(IMP),” where IMP stands for Idle [insertion] Mapping
Procedure. This is conceptually similar to ODUflex(GFP) in which the unused
payload bandwidth of the fixed-rate ODUflex(GFP) signal was filled with GFP Idle
frames. Note that IEEE 802.3 specifies inserting Idle characters only between
Ethernet frames, in the inter-frame gap (IFG). Hence, while these mappings are
Ethernet stream mappings, they are pseudo-packet aware rather than simple
CBR mappings.
20
then multiplexed into a higher rate OPUk. One consequence of the decision not
to allow switching ODUC signals is that no clients are mapped directly into an
OPUCn. All client signals must first be mapped into a legacy ODUj (j = 2, 3, 4,
flex) signal, which is then multiplexed into the OPUCn.
All ODUk signals carrying client information are multiplexed into the OPUCn
using the Generic Mapping Procedure (GMP) for rate compensation between the
ODUk rate and the aggregate rate of the OPUCn TS set that it occupies. GMP
was introduced with the previous generation of OTN. Here, GMP works by having
the source node, once per OPUCn multiframe, transmit a count of the number
of ODUk data words that it will map into that TS (or TS set) during the next
multiframe. The word locations that are not filled with the ODUk data are filled
with pad words, and the location of the pad words is determined based on
modulo arithmetic involving the count value.
21
before it gets scrambled for transmission. FEC has proved to be efficient in
correcting a very high number of errors in transmission due to noise or other
impairments present in high-capacity transmissions. The standard FEC uses a
Reed-Solomon RS (255/239) coding technique, in which 239 bytes are required
to compute a 16-byte parity check. Allowing service providers to extend the
distance between optical repeaters, FEC helps reduce both capital and
operational expenses while simplifying the network topography by being able to
skip amplifier sites.
IEEE 802.3 defined a relatively strong Reed-Solomon FEC for use with Ethernet
signals having rates ≥100 Gbit/s. The RS(544,514) “KP4” code covers blocks of
514 10-bit symbols in each FEC codeword, and can correct up to 15 errored
symbols per codeword. The overall performance is effectively comparable to the
RS (255, 239) “GFEC” code defined for legacy OTN, which was based on 8-bit
symbols. However, the RS (544, 514) achieves its performance with lower coding
overhead (544/514 = 1.0584 and 255/239 = 1.0669). This reduced FEC overhead
ratio is very similar to the rate ratio between an ODUC1 and a legacy OTU4 using
GFEC. Consequently, applying the KP4 FEC to an ODUC1 made it possible to
achieve a bit rate within <5 ppm of the OTU4 rate, which allows using existing
100 GbE/OTU4 PHY modules. Hence, a flexible interface for B100G based on the
RS (544, 514) FEC was ideal in terms of both the ability to reuse existing optical
modules and reusing Ethernet KP4 FEC IP.
22
carryMAC flows that are higher rate than a single PMD. The primary motivation
for developing FlexE was to provide flexible interface rates for data center
interconnections.
A FlexE client signal is an Ethernet MAC flow that is a CBR-type signal. The TDM
structure that allows multiplexing the FlexE clients is based on periodic “calendar
slots” that can be filled with a 64B/66B character from the MAC flow assigned
to that calendar slot. The frame format for each PHY consists of a slot carrying
the FlexE signal overhead followed by 1023 sets of 20 calendar slots for carrying
FlexE client data. The calendar associated with each individual PHY is called a
sub-calendar of the overall calendar for the FlexE Group. A FlexE client MAC flow
can own any number of calendar slots, and these calendar slots can either go
over the same PMD or be spread across multiple PMDs. As noted above, the
nominal rate of each calendar slot is 5 Gbit/s (corresponding to 20 calendar slots
per 100 Gbit/s PMD).
In addition to using FlexE for mapping packet data clients into the OPUCn, as
described above, OTN B100G has flexible options for carrying FlexE signals.
Specifically, when the client signal originates as a FlexE signal, there are three
options for mapping it into an OPUCn. One option is to terminate the FlexE signal
and map the FlexE client flows into the OPUCn as individual clients. A new FlexE
signal can then be created at the OTN demapper. The second option is called
“FlexE unaware” because it treats each 100 Gbit/s FlexE PMD stream as if it is
an individual 100 GbE signal. Hence, the mapper and demapper have no need
to know that a FlexE signal is present.
The third option for mapping a FlexE signal into an OPUCn is called “FlexE
aware.” One of the features of FlexE results from each FlexE client flow being
assigned only the calendars slots required for that client flow’s bandwidth.
Consequently, it is possible that the bandwidth of a PMD is not fully utilized. If it
is known that a PMD will remain only partially utilized, the unneeded calendar
23
slots are grouped at the end of the calendar and flagged as being
“Unavailable.” The PMD can then remove these calendar slots and transmit at
a corresponding lower (“crunched”) CBR PHY rate in order to reduce system
power, achieve longer signal reach and/or improve error performance. When a
FlexE-aware OTN mapper receives either a crunched FlexE signal or a full-rate
FlexE signal with Unavailable calendar slots, it will map the lower rate CBR stream
corresponding to the non-Unavailable calendar slots. Consequently, the ODUflex
carrying the FlexE client will require one fewer OPUCn TS for each Unavailable
FlexE calendar slot that is removed, thus making more efficient use of the OPUCn
capacity. The resulting signal is mapped into an OPUflex using the special bit-
synchronous mode of GMP (BGMP). BGMP generates the GMP count values
deterministically rather than deriving them from the input client rate and buffer
fill.
24
associated FEC. This separation allows creating PHY formats optimized for
different applications that are independent of the B100G signal format except for
its rate. The first PHY interface defined for OTN B100G signals was conceptually
inspired in part by FlexE. It is called Flexible OTN (FlexO) and is defined in ITU-T
Recommendation G.709.1.
Like FlexE, FlexO is a modular interface consisting of a set of 100 Gbit/s optical
PHY streams that are bonded together to carry an OTUCn. This allows using any
value of “n” for an OTUCn interface rather than defining only certain discrete
values of n (e.g., just n = 4 and 10). Since the ODUCn signal is constructed from
n of the 100 Gbit/s ODUC slices, it is well suited to using an n × 100 Gbit/s modular
PHY, with each 100 Gbit/s PHY carrying an OTUC slice. Note that if a set of m 100
Gbit/s PHYs are available, a subset of n PHYs can be chosen to carry an OTUCn
(n < m). For example, this would allow choosing the subset of PHYs with the best
optical channel characteristics, or carrying multiple OTUCn signals over a set of
PHYs. Since the initial version of FlexO is intended for client side interfaces, it was
ideal from a component cost and availability standpoint to reuse the existing
pluggable optical PHY modules developed for 100 GbE and OTU4. FlexO makes
partial reuse of the lane architecture and FEC structure from 100 GbE and 400
GbE Ethernet in order to leverage Ethernet IP.
25
Figure 8: Data flow from input client signal to optical line signal
16.0 Challenges
B100G also faces technical challenges such as channel modulation, line
transmission and product integration and packaging.
26
High-order QAM can increase spectral efficiency and transmission capacity, but
requires narrower linewidth lasers and better linear optoelectronic components.
Moreover, closer arrangement of constellation diagrams in high-order
modulation results in shorter transmission distance.
16.2 Line transmission
Compared with the ever-changing development of channel modulation, line
transmission has developed at a relatively slow pace. New fibers (such as G.654E)
with low loss and large effective area can reduce line loss and increase the
incident optical power, thus reducing the number of electrical repeaters. On the
equipment side, low-noise optical amplifiers are used to improve transmission
performance, and the commercial deployment of distributed Raman amplifiers
are accelerated. These technical means are adopted to ensure engineering safety
and reduce maintenance difficulties of Raman amplifiers and will play a more
significant role in extending transmission distance.
The OIF is developing IC-TROSA standards for packaging. Type-A standards are
specified for the SiPh technology that integrate modulators, drivers and receivers.
Type-B standards are defined for the InP technology, integrating lasers,
modulators, drivers and receivers. Leveraging the advantage of non-hermetic
27
packaging, SiPh enables BGA packaging, while the InP technology can integrate
optical components into a Goldbox.
18.0 Conclusion
Originally introduced in 2000, OTN has proven capable of continuing to evolve
and adapt to cover new applications and technology in the transport network.
The new OTUCn B100G standard is the next stage of that evolution. The modular
frame and rate structure readily allows:
1) A convenient way for the OTUCn signal to be divided for transmission over
multiple wavelengths.
2) Supporting multi-vendor interfaces of different rates.
3) A straightforward adaptation to carry future high-speed client signals, such as
Terabit/s Ethernet.
4) The introduction of FlexO for a modular interface that can exploit 100
GbE/OTU4 PHY modules.
28
While the initial application for OTUCn signals will be long haul connections, it
will see metro network applications as 400 GbE and FlexE become available and
data-center connectivity increases in importance.
With its mapping and rate flexibility, its associated FlexO PHY that takes
advantage of Ethernet modules, and its overhead optimized for service provider
transport networks, the B100G OTUCn preserves OTN’s role as the ideal
transport technology for service provider transport networks.
19.0 References
1. White paper on Moving to 100G and beyond by Sterling Perrin Senior
Analyst, Heavy Reading
2. The Evolution of ITU-T G.709 Optical Transport Networks (OTN) Beyond
100Gbit/s by Microsemi
3. ITU-T G.709/Y.1331 on Interfaces for Optical Transport Network
4. A Tutorial on ITU-T G.709 Optical Transport Network by Microsemi
5. Beyond 100G: The glass is heating up by Xiong Qianjin, Huwaei
Communications
6. Prospects of Beyond 100 G OTN by Shi Kai ZTE
7. Expert series Master ins and outs of OTN by Paul Littlewood, Fady Masoud
with Malcolm Loro
20.0 Glossary
ITU-T: International Telecommunications Union – Telecommunications
Section
B100G: Beyond 100G
OTN: Optical Transport Network
SONET: Synchronous Optical Network
SDH: Synchronous Digital Hierarchy
OTH: Optical Transport Hierarchy
ODU: Optical channel Data Unit
29
OPU: Optical channel Payload Unit
GFP: Generic Framing Procedure
OPU: Optical Payload Unit
ODU: Optical Data Unit
OTU: Optical channel Transport Unit
ODTUG: Optical Data Tributary Unit Group
InP: Indium Phosphide
BIP: Bit Interleaved Parity
Och: Optical Channel
FC: Fibre Channel
FEC: Forward Error Correction
ESCON: Enterprise Systems Connection
LAN PHY (10GBase R): LAN Physical interface
WAN PHY(10GBase-W): WAN Physical Interace
LO: Low Order
HO: High Order
GMP: Generic Mapping Procedure
PPM: Parts Per Million
IC-TROSA: Integrated Coherent Transmit & Receive Optical Sub Assembly
CBR: Constant Bit Rate
AMP: Asynchronous Mapping Procedure
MAN: Metropolitan Area Network
QAM: Quadrature Amplitude Modulation
30