Fixed Access Network Sharing - ACCESS AND HOME NETWORKING TECHNOLOGY STANDARDS - ASSIA Inc.
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
ACCESS AND HOME NETWORKING TECHNOLOGY STANDARDS
Fixed Access Network Sharing
Kenneth J. Kerpez, John M. Cioffi, Peter J. Silverman, Bruno Cornaglia, and Gavin Young
Abstract slice. The access node is generalized terminolo-
gy for a digital subscriber line access multiplex-
Deep fiber deployments for ultrafast broad- er (DSLAM), optical line terminal (OLT), cable
band are both technically and economically chal- modem termination system (CMTS), and so on.
lenging. Fixed access network sharing (FANS) The aggregation node aggregates traffic.
offers a highly enhanced form of virtual unbun- As fiber penetrates deeper, physical unbun-
dling of broadband access networks, enabling dling gives way to virtual unbundling, and FANS
cost sharing and the dynamics of a competitive provides the means to continue a vibrant compet-
landscape. Data, control, and management inter- itive ecosystem supporting economically viable,
faces are automated and harmonized among differentiated, and innovative services. FANS is
wholesale infrastructure providers and retail vir- synergistic with the software-defined access net-
tual network operators. Standardized FANS inter- work (SDAN) concept [5, 6].
faces can greatly decrease OPEX while increasing
customer satisfaction. When FANS is implement-
Competition
ed, network slicing logically partitions and isolates Current Multi-Operator Landscape
network resources among the VNOs, and such Competition today on copper infrastructure
sharing can split the cost of network upgrades is often via facilities-based physical unbundling,
among several entities. FANS works with virtu- with competitive providers deploying their own
alization, where control plane functions are access nodes and leasing copper loops from InPs.
migrated from dedicated network equipment into Physical unbundling allows a retail operator to
software running on commodity hardware, with essentially run all their own operations and inde-
FANS providing network as a service. This article pendently offer services. Physical unbundling is
describes the drivers behind FANS, several archi- very popular in some areas (e.g., Europe) and has,
tectures supporting FANS, and emerging stan- arguably, led to very low-cost and very high-speed
dards supporting FANS. service to consumers.
There is also some “virtual unbundling,” which
Introduction has two flavors: bitstream and virtual unbundled
Fixed access network sharing (FANS) applies local access (VULA). Bitstream is generally provid-
when an infrastructure provider (InP) controls ed by giving each VNO access to their customers
a physical access network that supports virtual at the IP layer and is simple resale of the service
unbundling to virtual network operators (VNOs) provided by the InP. Bitstream usually is offered in
[1]. FANS provides interfaces [2, 3] that allow a way that does not support differentiation of ser-
diagnostics and status data to be disseminated vices between VNOs. VULA is generally provided
from the InP to VNOs, and FANS allows a VNO at the Ethernet layer, and can enable layer 2 class
to request or perform changes in network con- of service differentiation and sometimes multicast.
figuration and control their own virtual network. Generally, the management, backhaul, and other
These FANS interfaces enable automated oper- operations are opaque to the VNO with current
ations spanning the InP and VNO domains. An virtual unbundling, although there are limited
objective of FANS is to enable VNOs to perform exceptions [7].
operations with virtual unbundling similar to their
operations with physical unbundling [4]. Fiber-Deep Ultrafast Broadband
Not only is the physical access network provid- Vectored VDSL2 and G.fast coordinate the sig-
ed by the InP and shared with FANS, but manage- nals across all the copper pairs emanating from
ment data and configuration are also shared. With an access node. Therefore, physical unbundling
FANS, network management is a shared respon- is technically unattractive with fiber to the node
sibility, with some management being performed (FTTN) using vectored very high rate DSL 2
by the InP and some by VNOs. There may be dif- (VDSL2), and with fiber to the distribution point
ferent levels of data sharing, distinguished by both (FTTdp) using G.fast because multi-operator vec-
the data shared between VNO and InP, and by toring is not yet standardized. Physical unbun-
the resolution and accuracy of the shared data. dling with fiber to the premises (FTTP)/fiber to
Figure 1 shows how a single network and its the home (FTTH) using passive optical network
equipment is administered by an InP, and a net- (PON) technologies could be possible using
work sharing system logically divides the physical an overlay network or separate wavelengths.
network resources between a number of VNOs, However, for all these new ultrafast broadband
each of which has their own virtual network deployments, physical unbundling is generally
Digital Object Identifier:
10.1109/MCOMSTD.2017.1600001ST Kenneth J. Kerpez, John M.. Cioffi, and Peter J. Silverman are with ASSIA Inc.; Bruno Cornaglia and Gavin Young are with Vodafone
1 ISSN 2471-2825/17/$25.00 © 2017 IEEE IEEE Communications Standards Magazine • March 2017economically unattractive. This is because phys-
ical unbundling would require many operators Access VNO 1
to each run fiber in the outside plant to each of node Aggregation
many small nodes, install all these small nodes, node
and deploy equipment at each of these small
nodes. Much of this fiber and equipment would VNO 2
be redundant with other operators’ deployments
and be underutilized.
The existence of multi-operator environments Shared network
will depend on virtual unbundling. Basic resale is InP domain VNO 3
insufficient; an ecosystem supporting innovative
service offerings from multiple operators depends Figure 1. Network sharing data plane, showing network slices in different
on a platform that allows retailers to perform colors.
operations in a way that is nearly indistinguishable
from physical unbundling. For example, FANS can Using an active sharing approach is more con-
enable multicast or real-time service variations venient than the passive one, mainly because of
such as turbo-boost, bandwidth reservation for the extra cost in the passive approach for each
video streams, and real-time charging. Moving operator to install and connect its own equip-
forward, FANS is especially desirable in its ability ment. On the other hand, active sharing without
to encourage such competitive innovation and using FANS limits the capability of the operator to
differentiation, and to drive economic growth of differentiate its offer from others, so active sharing
broadband services at lower cost and lower oper- can be less attractive. FANS extends active shar-
ational complexity. ing to encompass management and control func-
tions. With FANS the solution implemented is a
FANS Benefits “virtual unbundling,” so each operator has similar
Deploying superfast broadband networks is quite capability to implement its own access network,
costly in terms of both upfront investment and which is why FANS is an appealing option.
resources needed for design and implemen-
tation. Multiple parallel networks covering the Use Cases
same areas can lead to low take rate and long With current bitstream or VULA, operations “inter-
return on investment. Operators should therefore faces” between InP and VNO are often manual.
share their own infrastructure in order to reduce In conjunction with appropriate business arrange-
upfront investments and operational resources. ments, FANS can assist in automating operations
The International Telecommunication Union (ITU) interactions between InPs and VNOs; including
Commission for Sustainable Growth and the Euro- fault, configuration, performance monitoring, and
pean Commission have recognized that sharing optimization. In particular, standardized FANS
infrastructure can speed broadband rollouts. In interfaces allow efficient data exchange between
some countries, including Portugal, Spain, and the all parties. This can lower operation expenditure
United Kingdom, regulatory bodies have taken (OPEX) costs, improve customer satisfaction by
into account this consideration and have created enabling rapid response times, and increase the
rules for sharing existing passive infrastructures. number of customers.
Operators also create their own agreements for FANS can enhance competitiveness with com-
sharing the access network. peting broadband media. Multiple companies can
Two types of sharing are possible, as shown share the costs of infrastructure and its upgrades.
in Fig. 2: passive and active sharing. With pas- The InP can offer enhanced product offerings to
sive sharing, only the passive infrastructure is the VNOs, with enhanced value for providing
shared, including ducts, poles, and cabling; and access to automated data and control interfaces.
each operator installs its own access node equip- FANS use cases include:
ment with a fiber tree connecting it to the opti- Performance monitoring and optimization
cal distribution frame (ODF) in the central office is enabled. Each VNO can have automated real-
(CO). For PON, passive sharing leads to a cost time access to performance and fault monitoring
increase above a single infrastructure because data. This is useful for VNOs’ network monitoring
multiple parallel connection trees have to be built. as well as repair and troubleshooting operations.
In the case of active sharing, both passive infra- Customer relationship management (CRM) can
structure and access node equipment are shared; improve. VNOs can optimize the configurations
the drawback is that the VNO connects via bit- of their lines. Further, multi-line optimizations can
stream. Active sharing with bitstream has limited be performed across multiple operators’ lines by
customer management and service differentiation a centralized sharing system, which can increase
capabilities, which sometimes make active sharing performance of all lines [4].
unattractive. Passive sharing can give around 30 Fault correlation can be performed using
percent savings, while active sharing can lead up shared data to correlate multiple faults across mul-
to 40 percent savings, compared to deploying tiple lines and multiple service providers; and this
multiple parallel networks. can further be used to help coordinate dispatches.
To overcome this situation, active sharing with For example, pooled data across multiple VNOs
FANS uses VULA and virtualization concepts to and/or InPs can be used to identify a fault that
allow slicing the network elements so as to assign occurs in a single shared cable section. A single
dedicated resources to each operator. In this dispatch to fix that cable section is much better
case the cost of the passive infrastructure remains than dispatching to each troubled line separately.
the same, but new slicing capabilities have to be Configuration can be automated, so a VNO
developed. can rapidly turn up service. VNOs can offer ser-
IEEE Communications Standards Magazine • March 2017 2The centralized sharing Passive sharing
CO Shared Apartment/home
system may be adminis- fiber
Operator 1
tered by the InP, a VNO, or Shared Shared
Operator 1 fiber ODF
a third party. The central-
ODF
ized management system Shared
fiber
itself could be provided Operator 2
by multiple parties either Operator 2
via shared or open source
software, or via APIs Active sharing
between proprietary sys-
CO Shared Apartment/home
tems. With shared or open fiber
Operator 1 Shared Shared
source software, the cost Operator 1 fiber ODF
of the software is split
ODF Shared
across multiple parties. fiber
Operator 2
Area 1
Figure 2. Passive sharing and active sharing approaches, and their relative 10-year TCO per operator.
vices with different quality of service (QoS) levels, • Runs and redesigns customized systems in its
for example, to maximize speed or stability, or to own virtual network, such as diagnostics and
minimize delay or power usage. Data sharing can optimization and DSM/DLM systems [10]
help ensure that the lines can support the neces- • Provides specific and customized service
sary QoS attributes. through its own area network
Services innovation can flourish; for example, • Utilizes the network resources provided by
VNOs can provide different types of assured ser- the InP to provide its services
vices, business class services, sponsored services, • Obtains revenue by selling services to end
and so on. This can build on basic broadband users
offerings to grow the pie for all entities.
Network planning can be enhanced with net- Centralized Sharing System
work topology and capacity data. Lines in a small FANS is generally instantiated around a central-
geographic area should all have about the same ized sharing system, as shown in Fig. 3. The cen-
performance, so neighborhood data can be ano- tralized sharing system has southbound interfaces
nymized by a sharing system to provide line per- to equipment and northbound interfaces to VNO
formance projections to VNOs. systems. The centralized system performs certain
A single operator can exploit FANS for their functions, such as authentication, authorization,
company and segment their network among dif- and accounting (AAA), and arbitration of requests
ferent operation teams for business customers, for resources, data, and control. The centralized
consumer broadband, and/or mobile backhaul sharing system can perform operations such as
connectivity. Moreover, FANS can also facilitate diagnostics, configuration, and optimization for
mergers and acquisitions because operations can the VNO, as shown in Fig. 3 for VNO A; or the
initially be run separately but with an integrated external interface to the centralized sharing sys-
network. tem can enable these functions to be performed
by a VNO itself, which is the case for VNO B
FANS Components in Fig. 3. The centralized sharing system may be
Actors administered by the InP, a VNO, or a third party.
FANS defines interfaces between an InP (aka The centralized management system itself could
wholesaler) and multiple VNOs (aka retailers). be provided by multiple parties via either shared
The InP operates the physical network, while the or open source software, or application program-
VNOs interface to broadband consumers. ming interfaces (APIs) between proprietary sys-
The InP is responsible for deploying and man- tems. With shared or open source software, the
aging the physical network; the InP: cost of the software is split across multiple parties.
• Enables physical resource slicing and carries The centralized sharing system implements
out the slicing multi-tenancy, although the functionality may be
• Provides an interface to the VNO for data distributed among multiple systems or locations.
and control Aspects of FANS could be implemented without
• Gets revenue from resource leasing a centralized sharing system, for example, with a
The VNO leases resources from the InP, and distributed architecture simply having interfaces
the main VNO functions are: between the various players. In any event, the
• Operates, controls, and manages its own vir- centralized sharing system conceptually glues
tual area networks FANS together.
3 IEEE Communications Standards Magazine • March 2017VNO B
management systems
Diagnostics
VNO A Configuration
management systems
Optimization
External interface External interface
Standard sharing interface Standard sharing interface
Hypervisor
External interface External interface
Monitoring, Diagnostics
performance, AAA
Configuration Control
diagnostic, Resource management Central and
status, and Optimization
database configuration
other state data
•••
Equipment
interfaces Centralized sharing
system
Aggx
BNGx
Aggx
ANx
Aggz Aggz BNGz
Aggy Internet
Aggy BNGy
ANz
ANy
Figure 3. Centralized sharing system.
The components in Fig. 3 are part of the infra- • The business support system (BSS) inter-
structure that supports FANS: face supports ordering and billing functions
• The “hypervisor” manages authentication between InP and VNO. The BSS includes a
and authorization to use the centralized catalog of offers and resources, an ordering
sharing system, and manages resources. The interface, inventory (of physical and virtu-
hypervisor here oversees the virtualization al components, equipment, outside plant,
or slicing of the network, not the compute admissible configuration settings, etc.),
infrastructure. and revenue. The resource catalog should
• AAA: Authentication, authorization, and include physical resources, virtual resources,
accounting verifies user credentials, admits and services.
requests and limits access, and maintains • Test and diagnostics interface. This invokes
transactional records for billing and other specific test actions to provide diagnostics
purposes. data beyond what the equipment can pro-
• Resources management includes assignment vide during normal operation. It provides
of network bandwidth, equipment interfac- outside plant and network test data, and may
es, equipment computational resources, and provide analyses of these data. Test may be
management interface bandwidth/frequency abstracted to hide details of the network
of admissible requests. equipment capabilities and to provide results
• Southbound equipment management inter- in terms of the virtual services provided. The
faces typically utilize the existing manage- interface may issue commands for request-
ment capabilities of network elements or ing tests.
their management systems. • Logical inventory. This includes equipment,
• The northbound sharing interface should be interfaces, virtual ports and their assign-
standardized. This interface provides moni- ments, equipment configuration settings, and
toring, performance, diagnostics, status, and virtual functions inventory.
other state information from the network to Messaging across the southbound interface
the VNOs, and also relays or interprets con- from the centralized sharing system may be imple-
trol and configuration requests from VNOs mented by an abstraction layer or an adaptation
to the network. layer. An abstraction layer hides the details of
There are additional interfaces between InPs equipment interfaces to present a simplified inter-
and VNOs that may or may not involve the cen- face toward management systems. An adaptation
tralized sharing system: layer directly translates signals from one format to
IEEE Communications Standards Magazine • March 2017 4another format, and usually has a different adapt- The management system supports multi-tenancy,
Management system shar- er for each type of interface. where each VNO is a separate tenant.
ing separates the manage- Management system sharing separates the
Resources Management and Security management plane from the data plane, with
ment plane from the data
Resources must be assigned carefully, to control sharing and network slicing performed by the
plane, with sharing and access permissions, arbitrate conflicts, ensure cor- management systems. The data plane can remain
network slicing performed rect or fair resource utilization, and guarantee reli- unchanged, and data-plane functions such as
ability for the underlying physical infrastructure. packet forwarding continue to be performed in
by the management Resources need to be assigned, with data access the network elements. Aspects of the control
systems. The data plane and control separated, for the access network, plane may also support sharing and network slic-
can remain unchanged, equipment, and computing infrastructure. A VNO ing functions.
cannot be allowed to access private data about The centralized management system could
and data-plane functions another VNO’s customers. Sharing of resources provide functionalities that include the following:
such as packet forwarding must be managed to ensure that resources are • Security, which includes authentication to
properly allocated among the competing VNOs, verify user credentials, authorization to admit
continue to be performed and that any particular VNO cannot either impair requests and limit access, and accounting to
in the network elements. another VNO’s service. maintain transactional records for billing and
Aspects of the control Computing resources, including CPU, memo- other purposes
ry, and virtual network, can be shared between • Fault correlation, particularly for faults that
plane may also support any or all of the actors. Managing computing occur on lines or equipment that impact
sharing and network slic- infrastructure resources is particularly important multiple VNOs
for virtual functions. • Inventory maintenance of the physical plant
ing functions. and equipment, as well as the virtual assign-
Backhaul ment of resources
The backhaul network extends from the access • Data maintenance needed to access VNOs
node to whatever point the traffic is handed off to and equipment such as addressing
a VNO or service. Traffic on the backhaul broad- • Support of an automated data clearinghouse
band network needs to be segregated between that allows automated operations
different operators and services. Backhaul segre- • Providing data to assist VNOs with network
gation can use VLANs, MPLS tunnels, or software planning and to assist in development of
defined networking (SDN) approaches. An inter- innovative services and differentiated ser-
esting new VLAN approach is to use a new “oper- vices
ator VLAN” (O-VLAN) tag, which may be a third • Multi-line optimization across multiple VNOs
VLAN tag in addition to C-VLAN and S-VLAN tags Management system sharing allows a VNO
of IEEE 802.1ad Q-in-Q. This allows the VNO to to choose to perform the following operations,
manage two levels of VLANs (S+C VLAN) for its among others:
service configurations, while the InP only assigns • Services provisioning
the O-Tag for each operator. • Fault and performance management
A VNO may wish to use its own backhaul • Configuration of the network elements
network. In this case, data traffic may be handed • Testing and gathering of diagnostic data
off from InP to VNO at various reference points, • Line optimization
including the V-interface to the regional broad- • Call center operations to answer trouble calls
band network or the A10 interface to service pro- Unlike virtual node sharing, management-sys-
vider network(s) as defined in TR-101 [8]. The tem-based sharing can be implemented with cur-
handoff could be at a broadband network gate- rently deployed equipment.
way (BNG), in the aggregation network, or even
in the outside plant at a cabinet location. Virtual Node Sharing
Virtual node sharing is based on the concept of
FANS Architectures equipment slicing. Virtual node sharing is per-
Two types of virtualization are emerging [1]: formed within equipment, including access nodes,
• Equipment slicing, where network resourc- aggregation nodes, and virtual port mappers. This
es are virtualized with slicing to support may require the equipment to host a common
multi-tenancy. Separate tenants perform sep- execution environment for sharing, such as a
arate functions on logically separate parts segmented space running a version of Linux. A
of the equipment. This is most aligned with hypervisor controls the life cycle and resources of
virtual node sharing. virtual machines (VMs). Equipment slicing allows
• Full virtualization, where network functions interchangeable functions to be hosted similar to
migrate from equipment to being hosted on the way a data center can host virtual network
cloud infrastructure platforms. This is most functions (VNFs). This concept is part of the virtu-
aligned with management-system-based shar- al OLT (vOLT) in the Central Office Re-architect-
ing. ed as a Datacenter (CORD) initiative. FANS can
be considered a use case of CORD.
Management-System-Based Sharing The virtual access node model performs equip-
The centralized sharing system in Fig. 3 is the ment slicing on physical access nodes to abstract
core of management-system-based sharing. With them into multiple virtual access nodes, where
this sharing technique, a management system each VNO accesses a logically separate virtual
performs the network slicing at the management access node. Separate access node functions can
system level and not directly in the equipment be sliced independently, and as shown in Fig. 4
itself. The management could be virtualized and some functions may also be fully virtualized and
hosted in the cloud or at other operator locations. hosted on cloud virtualization infrastructure.
5 IEEE Communications Standards Magazine • March 2017SDN controllers for the
Virtual access Virtual access Virtual access Virtualized node access network can also
node (VNOx) node (VNOy) node (VNOz) functions
on top of the equipment be virtualized. There may
Access node (InP) be a hierarchy of SDN
controllers, with an end-
Access node (InP) FANS to-end SDN controller on
top of domain-specific
SDN controllers. Network
Function-specific equipment Virtual access Virtual access Virtual access
node (VNOx) node (VNOy) node (VNOz) Virtualized node SDN control can be divided
functions between the InP, having an
on top of high-volume
High-volume server server platforms infrastructure controller,
and the VNOs, having vir-
Figure 4. Deployment scenarios for virtual access node functions. tual controllers.
Some functions may also be fully virtualized, as functions and other VNFs and network services
shown in the bottom half of Fig. 4. VNFs imple- (NS); this can be thought of as fixed access net-
ment some of the functions that traditionally work as a service (NaaS). Here, the NFVI and
reside in the access node or the BNG. MANO support multi-tenancy, where each VNO
A virtual access node element represents the is a tenant, and they are logically separated. For
whole set of characteristics of a physical access a given function, each VNO would have separate
node. A centralized sharing system is still present VNFs, NSs, and VMs, thereby using the under-
and is involved as part of virtual node sharing. lying NFV components to allocate resources,
The centralized sharing system described earlier ensure privacy, perform life cycle management,
performs orchestration, and monitors and scales and so on.
virtualized and physical network resources. SDN controllers for the access network can
Other network nodes can similarly use virtual also be virtualized. There may be a hierarchy of
node sharing, including virtual aggregation nodes SDN controllers, with an end-to-end SDN con-
(e.g. Ethernet aggregation switch, multiprotocol troller on top of domain-specific SDN controllers.
label switching [MPLS] router, SDN switch) [1]. Network SDN control can be divided between
The port mapper concept is part of virtual the InP, having an infrastructure controller, and
node sharing. A port mapper maps a disparate the VNOs,a having virtual controllers.
set of physical ports into a logical set of ports The following are among the fixed access net-
assigned to each VNO. The port mapper is a work functions that may be virtualized:
virtual entity used to map logical ports over the • VLAN translation/addition/removal: The
host physical ports. The virtual ports are iden- access node would focus on basic con-
tified through virtual port IDs. For example, as nectivity, whereas additional VLAN tagging
shown in Fig. 5, an access node may assign each could be performed in the NFVI.
user-facing port to a separate VNO, and then the • Virtual inventory management performs
VNO references the port through its virtual port assignments of physical assets.
ID. The port mapper may be combined with a • Per subscriber QoS enforcement (e.g., polic-
virtual switch, which can intelligently forward data ing or shaping) enforces QoS policy, and
by inspecting packets before passing them on, allocates QoS and class of service (CoS) lev-
ensuring traffic isolation. The port mapper more- els.
over facilitates the customer migration as the end • Port-based access control/authentication is
customer maintains the same physical ID and only performed, for example, by using a central-
changes the virtual port ID when moving from ized 802.1x agent
one operator to another. • Traffic is managed, filteried, and shaped, and
flow is controlled.
Full virtualization • Forwarding, traffic steering, load balancing,
Full virtualization moves functions into cloud and SDN control are virtualized.
platforms/data centers. Virtualization generally • Application awareness, deep packet inspec-
follows the architecture of the European Telecom- tion (DPI), and services-aware networking
munications Standards Institute (ETSI) Industry are virtualized
Specification Group (ISG) for Network Functions • There can be virtual gateway functions, resi-
Virtualization (NFV). Here, the NFV infrastruc- dential and business.
ture (NFVI) runs a system such as OpenStack or • Control and configuration: Each VNO con-
CloudStack, and the VNFs run on this infrastruc- trols and configures their own virtual access
ture. Management and orchestration (MANO) node dataset of configuration objects.
are controlled by a system such as Open Platform • Diagnostics and state information: Each VNO
NFV (OPNFV), Open Source MANO (OSM), accesses virtual functions providing test,
Open Orchestrator (Open-O), or a proprietary diagnostic, performance, and status informa-
system. tion.
FANS with full virtualization can extend man- • Dynamic rate allocation (DRA): This function
agement-system-based sharing to include control controls traffic scheduling, such as dynami-
IEEE Communications Standards Magazine • March 2017 6Physical deployment OSS/NMS logic network view
ONU1 OLT
Slot 1
Physical port ID
P1 vAN1
CPE 1 P2 Virtual port ID
P3 P1
CPE 2 P4 CPE 1
Slot 2 P2
P3
P1 CPE 3 P4
P2
P3
OLT vAN2
P4
Virtual port ID
ONU2 P1
Physical port ID P2
CPE 2 P3
P1
CPE 3 Splitter P4
P2
P3 P5
P4 Slot 1 P6
P7
CPE 4 P8
Physical port ID
P9
P1 Physical port ID
P2
P3 Slot 1
P4
CPE 4
ONU3
InP port mapper
Management IP address
Figure 5. Port mapper [1].
cally varying the G.fast asymmetry ratio, or purpose of DSM in DSL environments. Sharing
configuring PON dynamic bandwidth alloca- data on cable plant and DSL configuration and
tion (DBA). performance allows DSM level 2 and 3 multi-line
• Dynamic resource assignment can be virtu- optimizations and DSM level 1/DLM single-line
alized (e.g., access and backhaul bandwidth optimizations [10] to enhance the performance
assignment). of all lines.
• There can be virtualized dynamic spectrum Figure 6, from ND1518, shows a somewhat
management (DSM) and dynamic line man- complicated view of data sharing interfaces,
agement (DLM) [10]. including interfaces for DSM/DLM, BSS/ordering,
• Power control entity (PCE), cross-layer plant inventory, and plant test and repair. Three
low-power mode control, for G.fast: There entities comprise the InP: two access node opera-
are a number of thresholds and other settings tors (ANOs), and a transmission path facility (TPF)
that can be varied to configure low-power provider.
mode on individual transceivers, and these
settings and primitives can be determined in Broadband Forum DSL Data Sharing
a virtualized power control entity and com- Broadband Forum TR-349 [3] describes and
municated to the transceivers. defines data sharing for managing DSL. DSL data
• VDSL/G.fast vectoring control and manage- sharing architectures are described, with cen-
ment: Virtualized functions can control part tralized and distributed architectures presented.
of the vectoring configuration, and could High-level use cases are defined and requirements
even calculate vectoring coefficients [9]. presented for these use cases. The data and con-
trol parameters applicable to each use case are
Standards identified, and most parameters are common to
An initial effort at standardizing part of FANS for all use cases. Two types of DSL data sharing inter-
DSL was undertaken in the U.K. Network Interfac- face are defined in detail: profile-level (general)
es Coordinating Committee (NICC) [4]. This then and parameter-level (specific). The parameter syn-
fed into the definition of an interface for DSL data tax is defined in YANG data models for G.fast and
sharing in the Broadband Forum [3], and now a for VDSL in Broadband Forum TR-355.
larger project on FANS is underway in the BBF
[1, 2] to define FANS architectures, requirements, Broadband Forum
interfaces, and so on. Fixed Access Network Sharing
The Broadband Forum has embarked on a pro-
NICC ND1518, Data Sharing for DSM gram to specify FANS. This is to investigate tech-
The NICC specification ND1518 [4] describes nical aspects associated with FANS that involve
the use of sharing data between operators for the the access network, including access nodes and
7 IEEE Communications Standards Magazine • March 2017aggregation nodes. The Broadband Forum FANS
project should generate several specifications, VNO A VNO B
the first of which is FANS — Architecture and DSM/DLM
Nodal Requirements [1]. This document identifies
architectures and interface points for FANS. Both
management-system-based sharing and virtual-ac-
cess-node-based sharing are included. Require- Centralised system
ments are included, and operation, administration, AAA
and maintenance (OAM) and other operations Data and control DSM/DLM
are also described.
A project has also been started in the Broad- Inventory DSM,
band Forum to specify FANS access network shar- MPF data
ing interfaces [2]. Further specifications, which
have not yet been started, include access network InP 1 InP 2 MPF provider
virtualization and SDN-enabled FANS. DSM/DLM Test
Summary Inventory and
DSM data
Inventory and
DSM data MPF data
Ultrafast fiber-deep broadband deployments are
changing the competitive landscape. With current Figure 6. Simplified view of data sharing [4].
bitstream or VULA virtual unbundling, operations
interfaces between InPs and VNOs are often not [3] Broadband Forum TR-349, “DSL Data Sharing,” 2016.
real time, and VNOs provide simple resale with lit- [4] NICC ND1518, “Data Sharing for DSM,” 2015.
tle or no differentiation. Sharing network resourc- [5] K. Kerpez et al., “Software-Defined Access Networks,” IEEE
Commun. Mag., vol. 52, no. 9, Sept. 2014, pp. 152–59.
es and management interfaces will allow virtual [6] K. Kerpez and G. Ginis, “Software-Defined Access Network
unbundling to be economically and operationally (SDAN),” CISS 2014: The 48th Annual Conf. Info. Sciences
efficient, and enable vibrant competition based and Systems, Mar. 19, 2014.
on differentiation of offered services between pro- [7] W. Schramm, “NGA Regulation in Austria,” TNO DSL Sem-
inar, June 2013.
viders. [8] Broadband Forum TR-101i2, “Migration to Ethernet Based
FANS extends virtual unbundling to unbundle DSL Aggregation,” 2011
management and control functions, with network [9] J. Cioffi, “FULLY Unbundled Vectored DSLs: It Is Absolutely
sharing and data sharing. FANS opens up man- Possible!” G.fast Summit May 2016.
[10] Broadband Forum TR-197i2, “DQS: DSL Quality Manage-
agement and control interfaces such that VNOs ment Techniques and Nomenclature,” 2014.
can perform the same operations as they would
with physical unbundling, where they own and Biographies
operate their own network elements. Much as the Kenneth J. Kerpez [F’04] (kkerpez@assia-inc.com) received his
Internet offers a platform for innovative applica- Ph.D. from Cornell University in 1989. He worked at Bellcore
tions, so FANS can offer a platform for innovative and Telcordia for 20 years, and he now works at ASSIA. He
became an IEEE Fellow in 2004 for his contributions to DSL
broadband services. technology and standards. He has many years of experience
Standardized interfaces and a central manage- working on networks of all sorts, including DSL, fiber access,
ment system are keys to enabling FANS in the home networks, wireless systems, Wi-Fi, broadband service
near term. In the longer term, virtual nodes, SDN assurance, IPTV, IP QoS, triple-play services, and virtualization.
control, and full virtualization will all feed into John M. Cioffi (jcioffi@assia-inc.com) received his B.S.in elec-
enabling FANS. Resource control, AAA, securi- trical engineering in 1978 from the University of Illinois, and his
ty, and configuration control must be carefully Ph.D. in electrical engineering in 1984 from Stanford. He was
administered with FANS to ensure privacy and with Bell Laboratories, 1978–1984; IBM Research, 1984–1986;
and has been a professor of electrical engineering at Stanford,
avoid harm to the network. 1986–present, now Emeritus. He founded Amati Com. Corp in
FANS offers many benefits, both to InPs and 1991 and was officer/director from 1991 to 1997. He currently
VNOs: is chairman and CEO of ASSIA, Inc. His specific interests are in
• Automated interfaces lower operational costs the area of high-performance digital transmission. He has been
the recipient of numerous highly prestigious awards, has pub-
relative to manual interfaces, both for the lished over 600 papers, and holds over 100 patents.
InP and the VNOs.
• Fault correlation across multiple operators’ Peter Silverman (psilverman@assia-inc.com) is director, Stan-
lines is enabled, again lowering operations dards and Technical Marketing at ASSIA Inc. Prior to employ-
ment at ASSIA he was employed at Bell Laboratories, Ameritech,
costs. 3Com, and Valo Inc. before taking his current position at ASSIA
• Multi-line, multi-operator optimizations are in 2005. He has edited numerous international telecommunica-
enabled, which increases performance of all tions standards, and is a co-author of two books, Understanding
lines. Digital Subscriber Line Technologies (Prentice-Hall, 1999) and
DSL Advances (Prentice-Hall, 2003) and 10 patents.
• Multiple companies can share the costs of
network upgrades to superfast broadband. Bruno Cornaglia (bruno.cornaglia@vodafone.com) received
• It enables enhanced service levels, services his Degree in electronic engineering from Politecnico di Torino
differentiation, and innovation. in 1990. He worked at CSELT, the research center of Telecom
Italia Group, Omnitel, which then became Vodafone Italy, and
• The InP can offer FANS as an enhanced ser- now at Vodafone Group Services. His main areas spread from
vice to the VNOs. microwaves, where he developed the concept and solution of
• The VNOs can offer enhanced services to microwave Ethernet with adaptive modulation;, satellite, where
the broadband customers (e.g., enterprises). he worked on defining new SDH payloads to be used in satellite
links; and now fixed broadband, including copper, cable, and
fiber technologies, where he is working on different innovative
References approaches, like distributed architecture for cable, NG-PON2 in
[1] Broadband Forum WT-370, “Fixed Access Network Sharing fiber, and full access virtualization. He is an author or co-author
— Architecture and Nodal Requirements.” of more than 20 papers.
[2] Broadband Forum WT-386, “FANS Access Network Sharing
Interfaces.” Gavin Young [M] (gavin.young2@vodafone.com) received his
IEEE Communications Standards Magazine • March 2017 8B.Sc in electrical and electronic engineering from the University of Wales, Swansea, in 1985 and his M.Sc. in communications engineering from the University of London (Imperial College of Science and Technology) in 1986. He is head of the Fixed Access Centre of Excellence within Vodafone, where his inter- ests include technology strategy, architecture, innovation, and performance of broadband networks spanning copper, cable, and fiber technologies. He is a Fellow of the IET and Distin- guished Fellow of the Broadband Forum. 9 IEEE Communications Standards Magazine • March 2017
You can also read
