Comparing 3 Network Upgrade Options for Advanced Multimedia Service Delivery
by: Amitkumar Mahadevan and Laurent Hendrichs
(This article originally ran in the April 2011 issue of OSP Magazine)
Video is changing everything. Service providers throughout the world, motivated equally by the need to diversify their revenue streams as well as maximize their return from capital investments, are faced with the difficult challenge of cost-effectively upgrading their existing networks while providing customers with the robust network required to handle the increasing bandwidth demands of Triple Play services, high-definition video, and other advanced multimedia services.
VDSL is the current access technology of choice for Triple Play services and is enabling continuous increases in throughput speeds as well as extending services to more homes. Recent technology improvements are also conquering some of the performance-inhibiting hurdles inherent in DSL technology. VDSL pair bonding increases the maximum attainable bit rate available to VDSL end-users. Pair bonding also increases the service area for a given deployment.
Additional performance gains can be achieved when vectoring is implemented, either alone or in tandem with pair bonding. Alone or combined, these techniques allow service providers to design and deploy multimedia service networks that incorporate fiber in their backbone while allowing them to use their existing copper infrastructure for last-mile connections to their customers. These hybrid networks deliver dramatically improved capacity, performance, and stability.
Hybrid Network Realities
The widespread popularity of broadband multimedia services is a double-edged sword for telecommunications service providers. Declining voice traffic revenues, rising consumer expectations, and fierce competition from cable operators make service bundles, in particular those that include high-definition television, an essential survival strategy. Delivering such video-intensive services requires provisioning of bit rates from 20 to 100 Mbps to the end user.
Until recently, conventional wisdom held that only a fiber-to-the-home (FTTH) network architecture could deliver this level of performance. And unfortunately the wholesale deployment of FTTH is so costly that it strains even the deep pockets of the world’s largest telecommunications service providers.
Figure 1 illustrates a cost-comparison of the estimated installation costs associated with fiber and DSL installations for several international markets. From the data presented, it is clear that completely replacing an existing network is extremely costly and time-consuming. As a result, most service providers are pursuing a hybrid fiber-to-the-node (FTTN) architecture which provides subscribers with copper-based Last Mile links and xDSL technology.
Figure 1. Cost Per Household for Broadband Deployment/Installation.
Using the installed base of copper local loops eliminates the costly process of pulling fiber through a neighborhood and bringing it to each home. Additional savings are realized because most xDSL customer premises equipment can be installed by the subscriber.
One such option, of course, is VDSL. Significant advances in the next-generation of DSL technology (i.e., VDSL,) have improved the reach, stability, and overall capacity of legacy copper infrastructures so that the deployment of FTTN networks that reliably deliver 20 to 100 Mbps is achievable at a fraction of the cost of a purely optical solution.
VDSL is key to the deployment of Triple Play services within a twisted-pair access network. The increased bandwidth provided by VDSL allows the deployment of a high-speed access channel for video applications over Internet protocol (IPTV) while supplying sufficient capacity for the simultaneous support of high-speed Internet access and Voice over Internet Protocol (VoIP) applications.
Though it offers quick and easy deployment of advanced broadband services, telco copper is limited by the laws of physics. The amount of bandwidth available to users is limited by their distance from the fiber termination point, quality of the wire, and the amount of crosstalk in the cable. Factors include:
- Increased signal power loss is introduced with increasing frequency and distance.
- Multi-pair cables are subject to both near-end and far-end crosstalk coupling that increases with increasing frequency, making higher frequency signals more susceptible to degradation from crosstalk.
In general, the combination of signal loss introduced by a twisted pair cable and crosstalk introduced by signals in adjacent wire pairs within the same cable limit the range at which high-speed VDSL services may be deployed. Therefore, for a specified service quality objective on a single access wire pair, service providers limit deployment to customers within a set maximum radius from the central office (CO) or remote terminal (RT).
Customers located close enough to the CO may be served directly by equipment located inside the CO. To reach customers located further away, the service provider may deploy a fiber fed cabinet that contains the VDSL equipment serving the customers located within a maximum serving radius from the cabinet.
When additional bandwidth and increased distance are required, pair bonding can be incorporated to support another data channel over a second twisted-pair found in the Last Mile runs of most residences and commercial installations. Bonding logically combines the capacity of the 2 channels in a transparent manner so the subscriber sees a single connection that is theoretically capable of delivering up to 200 Mbps in performance.
Depending on the cabinet’s location, there may be a certain percentage of end-users whose access line(s) fall outside the maximum reach limit for service deployment on a single wire pair. Customers located outside this serving radius will likely not receive the objective service quality on a single pair due to large loss and possible excessive level of crosstalk in the long loop. For customers having more than 1 wire pair available to the home, pair bonding may be used to provide the service with extended reach, where the bit rates on each loop are lower than the objective rate but the combined bit rate exceeds the objective rate for the service.
With proper understanding of the rate versus reach tradeoffs, bonding may be used to extend the reach of basic xDSL service to customers outside the single wire pair serving radius, or to provide higher capacity to end users than on a single xDSL access link within the single pair serving radius.
Service providers today can employ Ethernet pair bonding to help their rate and reach challenges in a hybrid network. The sidebar “Pair Bonding Uncovered” provides a very brief overview of the Ethernet pair bonding mechanism defined in ITU-T Recommendation G.998.2. (For complete information, refer to the complete G.998.2 Ethernet Bonding Recommendation [a], and the Ethernet in the First Mile Standard [b].)
Use Case: Analyzing 3 Options for Broadband Network Upgrades
The use case shared below, referred to as W-Telecom network, illustrates how service providers today may determine which choice is best to upgrade their legacy network. The study examines the implications of each of 3 Last Mile upgrade options when upgrading legacy customers to Triple Play service rates:
- Passive Optical Network (PON)
- VDSL pair bonding
Figure 2 illustrates the 4 classical architectures of Last Mile access technologies based on fiber and copper.
Figure 2. Last Mile Access Options.
W-Telecom’s Network in 2010
In this case, Figure 3 illustrates graphically the relative distribution of W-Telecom’s current Last Mile customer deployments.
From Figure 3, it is seen that approximately:
Figure 3. W-Telecom Last Mile Customer Deployments in 2010.
- One-third of its deployments is FTTB: These deployments, even when installed post-construction, appears to be cost-effective due to multiple customers being serviced by a single fiber installation/deployment.
- One-fifth of its deployments is FTTH: These deployments are primarily found in neighborhoods and buildings where fiber was installed as part of the initial construction.
- One-half of its deployments is a combination of VDSL and legacy ADSL where most ADSL customers that are geographically located close enough to the CO/Cabinet have already been upgraded to VDSL service. The remaining legacy ADSL customers are primarily single family homes located too far from a DSLAM to be upgraded to VDSL.
- These distribution of customers per network are detailed in Figure 4.
Figure 4. W-Telecom Customer Deployment Information.
W-Telecom’s plan to cost-effectively upgrade their legacy ADSL deployments to allow Triple Play service rates is driven by the following factors:
- The bandwidth available to legacy ADSL customers is too low for high end services (e.g., IPTV), resulting in low average revenue per user (ARPU).
- Upgrading legacy ADSL customers to a guaranteed 30 Mbps service could increase ARPU by $15, thus generating potential $250 million of revenue annually.
- The operator’s technology upgrade requirements include: Maximizing the return on investment, leveraging the existing infrastructure, and their upgrade must be field proven.
Last Mile Technologies — Upgrade Analysis
Option A: PON Upgrade
Despite equipment cost reductions and improvements in installation time, fiber installation costs continue to remain above $1,000 (USD) in the majority of industrialized economies. While PON is an attractive option for a long-term service upgrades, the investment per subscriber remains extremely high.
Figure 5 compares the installation cost per subscriber, in U.S. Dollars, for Europe, the U.S., and Japan.
Figure 5. Typical Cost for Fiber (U.S. Dollars per Subscriber).
Given the costs stated previously, it is unrealistic to think that W-Telecom would upgrade their legacy ADSL customers to PON service on a mass-market basis.
Option B: VDSL Upgrade
Figure 6 shows VDSL as a strong alternative. In fact, 30/5 Mbps service is achievable for a maximum loop length of 3000 ft (914 m) under the following conditions:
Figure 6. Rate versus Reach for VDSL Profile 17a; 12 Self-Crosstalk Disturbers; 26 AWG; UPBO Off.
- Profile 17a (17 MHz bandwidth)
- 12 Self-Crosstalk disturbers
- 26 AWG wire
- Upstream Power BackOff (UPBO) is turned off.
Option C: VDSL Pair Bonding Upgrade
Figure 7 shows that pair bonding increases VDSL’s performance dramatically. In fact, 30/3 Mbps service is achievable for a maximum loop length of 5100 feet (1550 m) under the following conditions:
Figure 7. Rate versus Reach for VDSL Pair Bonding; 24 Self-Crosstalk Disturbers; 26 AWG; UPBO Off.
- 24 Self-Crosstalk disturbers
- 26 AWG wire
- UPBO is off.
In this case study, W-Telecom must continue to use a hybrid network but can now evaluate the advantages and disadvantages of each of the three upgrade options. Several advantages and disadvantages for each are detailed below.
Advantages: Easily achieves bandwidths in excess of 30 Mbps; Proven technology already deployed by the operator. Disadvantages: Rising installation cost as the majority of new deployments shifts from pre-wired homes to green field.
Advantages: Lower cost solution than PON; Proven technology already deployed by the operator. Disadvantages: Requires investment in new cabinets in order to install DSLAMs close enough to subscribers for an upgrade to VDSL rates.
VDSL Pair Bonding
Advantages: Doubles rates at any given loop length; Allows reach extension of ~2000 ft for typical VDSL rates; Innovative technology currently in field trials at multiple operators around the world; Disadvantages: Requires at least two twisted pair running to the home. Not applicable where only single pair are used.
When comparing the 3 upgrade options side-by-side, 2 conclusions are immediately obvious:
- PON is an attractive option for a long-term service upgrade. However, the investment per subscriber remains extremely high.
- Despite the need for 2 VDSL ports, the increased service radius makes VDSL pair bonding an economical alternative to upgrade legacy ADSL subscribers.
General findings of this case study also include:
Fiber remains 3- to 4-times more expensive than VDSL for service rates of approximately 30 Mbps.
VDSL pair bonding offers significant advantages over VDSL and xPON, including: Second line costs are more than offset by the savings realized due to the increased service radius of each cabinet/central office; VDSL pair bonding is the most economical technology currently available when upgrading subscribers located outside of the single-pair VDSL service radius; Improved quality of experience (QoE) resulting from the additional robustness provided by the second VDSL pair.
There is a minimum lag between investment and revenue curves resulting in a significantly reduced break-even point; Investment costs are primarily driven by CPE upgrades and installation.
Continuous technology innovations extend the capabilities of the existing infrastructure to enable advanced broadband services. For example: VDSL pair bonding extends the attainable rate/reach using the existing wired infrastructure; Vectored VDSL will enable further improvements by virtually eliminating crosstalk interference; VDSL pair bonding combined with vectored VDSL will enable rates of 100 Mbps out to 3000 ft (900 m.)
The Legacy Plant — A Valued Asset
When a second wire pair is available to an end-user, pair-bonded VDSL has been shown to be a viable solution for providing extended reach or higher bit rate over deployment of VDSL service on a single pair. Pair-bonded VDSL provides rate and reach gains over single pair VDSL systems with profiles 8, 12a, 17a, or 30a. Figure 6 and Figure 7 provide summaries of rate and reach gains for common service deployment scenarios. It should be noted that extended rate and reach results in greatly reduced capital expenditures thanks to a reduction in the number of cabinets required to service a given area.
The lower costs and advanced capabilities of hybrid copper/optical systems are allowing many major telecommunications providers to push forward with their multimedia network upgrade roll outs, despite reduced access to capital and slimmer profit margins.
Many service providers once considered staunch proponents of FTTH architectures have reconsidered their commitment and are moving away from the FTTH option instead for the economic realities and price/performance advantages of VDSL.
Bonding technology, offered by Ikanos as an example, enables DSL service providers to overcome bandwidth limitations inherent to copper infrastructures. VDSL pair bonding increases the capacity of subscriber lines and enables service providers to dramatically increase throughput speeds and extend broadband services to more homes.
Further advances in VDSL technology are on the horizon. Vectored VDSL is expected to enable fiber-class performance allowing service providers to deliver advanced multimedia services at a fraction of the cost of other technologies such as cable and FTTH. Alone or combined, these technologies allow service providers to maximize their investment in copper infrastructure while efficiently deploying the bandwidth necessary to deliver advanced multimedia services to customers.
About the Authors
Amitkumar Mahadevan is senior staff engineer at Ikanos; he has more than 10 years of experience in the broadband industry. Laurent Hendrichs is product marketing manager at Ikanos; he has more than 10 years of experience in the broadband industry. For more information, please email:email@example.com or visit www.ikanos.com.
a. ITU-T Recommendation G.998.2, “Ethernet-based multi-pair bonding,” January 2005.
- IEEE 802.3ah (2004), “Carrier Sense Multiple Access with Collision Detection (CSMA/CD) access method and physical layer specification – Amendment: Media Access Control Parameters, Physical Layers, and Management Parameters for Subscriber Access Networks.”
- Refer to the white paper titled: Enabling Network Upgrades for Advanced Multimedia Service Delivery – VDSL Pair Bonding Extends Broadband Delivery and Increases Speeds for Advanced Multimedia Service Delivery by Massimo Sorbara, Amitkumar Mahadevan, and Laurent Hendrichs. The section titled Performance Gains with Bonding, provides additional information including the specifications outlined in the above excerpt. August, 2010.
Pair Bonding Uncovered
The standards-based approach to Ethernet bonding is applicable to any physical layer transport type (e.g., VDSL2, ADSL2, SHDSL, etc.). To the upper layers in the protocol stack, the bonded links appear as one single high-speed link transporting Ethernet packets.
Figure 1 provides a functional block diagram of the bonding process at the transmitter. The bonding block receives Ethernet packets from the Ethernet layer above. The bonding block first removes any overhead associated with the Ethernet packet, specifically any inter-packet gap (IPG) bytes and the preamble. The packet segmentation block takes the resulting Ethernet data frame and breaks it up into smaller data fragments.
Figure 1. Functional Block Diagram of Ethernet Bonding.
Two bytes of overhead are added to each data fragment, and they are then distributed among the N lines in the bonded group for simultaneous transmission to the far-end. The bonding receive function at the far-end receives the fragments on all of the lines in the bonded group and then reassembles the original packets based on the information received in the overhead bytes.
Figure 2 shows a diagram of the fragmentation process. Only the Ethernet data frame is broken up into smaller data fragments. Two bytes of overhead are added to each fragment:
Figure 2. Fragmentation of Ethernet Packets for Bonding.
- 14 bit sequence identifier (SID);
- One bit for a start of packet (SOP) indication;
- One bit for an end of packet (EOP) indication.
The minimum data fragment size is 64 bytes and the maximum fragment size is 512 bytes.
When transmitted on a line using VDSL, the packet transfer mode transmission convergence (PTM-TC) layer function transports the bonding fragments by adding one byte of overhead for every 64 bytes of data received from the bonding layer. This PTM-TC function is referred to as 64/65-octet encoding, which is used to convert the data blocks received from the bonding layer into a bit stream (and resulting bit rate) for transmission using VDSL. Also, at the end of each data block, the 64/65-octet encoder appends 2 or 4 bytes of CRC (i.e., a frame check sequence [FCS],) for the detection of any errors in the received sequence.
Recommendation G.998.2 and IEEE 802.3ah define a differential delay requirement of 15,000 bit times between any 2 lines in a bonded group, and a maximum bit rate differential of 4:1 between the highest and lowest bit rates in the bonded group.
Amendment 1 to G.998.2 has updated the differential delay requirements when VDSL transceivers are used on each link to accommodate for the jitter effects introduced by various transceiver elements, namely transceiver jitter caused by transmitter and receiver buffering, symbol rate, sync symbols, and forward error correction (FEC). With this amendment, a two-pair bonding group transmitting 100 Mbps net data rate at a 4:1 ratio would need to support a differential delay of 65,000 bit times.