DSL Rings White Paper

Transcription

1 INTRODUCTION DSL Rings (DSLR) is a patent pending access technology that re uses existing copper telephone cabling to provide bandwidth of up to 400 Mb/s with Quality of Service (QoS) and Efficient Multicast. Unlike fibre or cable based alternatives DSLR can also provide E911/E999 service when the power is out. DSLRWP Genesis Technical Systems 1

2 ANALYSIS Telecom Service Providers are being forced to examine their network s last mile technology due to a number of factors, including VoIP based competition from cable providers; and whilst video offerings have been in their plans for many years they have had limited success. Current implementations of copper telephone networks cannot support the bandwidth necessary to offer a realistic streaming Standard Definition Video (SDV) service; even with MPEG4 compression (requires approximately 4 Mb/s guaranteed/channel). VDSL2 technology has increased the bandwidth carrying capability of the wires significantly but, as with all DSL based technologies, is highly distance from the DSLAM sensitive (see Figure 2). Other technological alternatives exist but they all require significant financial investment to upgrade the physical cabling, generally to fibre based technologies. The business case for these deployments varies from situation to situation but certainly does not apply to all cases. Any additional fibre deployment in the access network, other than for green field situations, cannibalises the existing copper wire infrastructure to varying degrees. FTTP/H replaces the entire copper wire infrastructure; and FTTC & FTTN requires replacement of a substantial amount of existing, invested in, revenue generating infrastructure. Wireless based network upgrades, whilst being relatively inexpensive compared to fibre, often require spectrum licensing, have physical security challenges, and peak bandwidth difficulties. They are also subject to disruption, location and reception issues, and they lower the barrier to market entry as competing carriers can deploy such solutions almost as easily as incumbents. Both wireless based and fibre based bandwidth enhancement options result in cannibalising the existing copper wire that is the single most valuable asset to the telcos. To summarise, these developments don t meet market demands, require disproportionate investment, or reduce competitive differentiation. One alternative is shared bandwidth. SHARED BANDWIDTH Shared network bandwidth has always been a part of our telecom experience. Statistical multiplexing is used by every carrier on earth though the ratios vary from carrier to carrier and location to location. DSLRWP Genesis Technical Systems 2

3 Cable networks are entirely based on bandwidth sharing via a bussed model (i.e., one common physical wire that everyone connects to). This means that everyone on that network segment can see (if they have sufficient equipment and understanding) what everyone else is doing on that cable. The point is that bandwidth is currently a shared resource in our networks, it has always been a shared resource (at varying levels) and there are economic benefits to retaining a shared bandwidth model. Bonding enables a significant increase in bandwidth at virtually every distance provided, as long as there are pairs available for the purpose. Bonding is when the original high bandwidth signal is split into several pieces and then each uses several different physical pairs as a single transmission link. This process is specified under the moniker G.Bond (ITU G.998.1, 2, 3). The difficulty of applying bonding by itself to the existing cable plant is that there are generally between 2 and 4 pairs going into each residence. If the current standard of 4 Mb/s is used, this yields a maximum of 16 Mb/s available to each house (this bandwidth is still shared once it hits the DSLAM). Typically there are only 2 pairs in residences and the maximum bandwidth would be 8 Mb/s. This is still considered to be very tight for video transmission, even with MPEG4 compression (which is not very common) as the bandwidth and latency needs to be guaranteed. DISTANCE REDUCTION A frequently used technique to reduce the transmission distance seen by electrical (and optical) signals is to regenerate the signal somewhere along the path. This has the effect of resetting the transmission distance to zero and starting again. From the perspective of the telecom network, simply regenerating DSL signals serves little purpose and provides no real value add. For example, to maintain maximum VDSL2 bandwidth (at a 500 ft distance) would require digging up all the large >1000 pair cables every 500 ft to deploy a regenerator; not to mention all the smaller cables as well. A solution used in the optical domain was to turn the regenerator site into an Add Drop Multiplexer (ADM). This allowed the main signal to be completely regenerated, and at the same time, add and drop traffic at that point. The ADM configuration was later modified so that the ADMs could be arranged in closed rings so that all bandwidth was available to all sites if they needed it. Previously it had been put in straight lines and wasted large amounts of bandwidth as the first position in the line had access to the whole bandwidth but rarely needed it, the next position typically didn t add too much on top of the first, and so on. Gateway nodes were inserted in the ring so that connections to the main network were maintained. This configuration describes a collector ring and is now widely used in metro level optical networks. DSLRWP Genesis Technical Systems 3

4 APPLICATION TO CURRENT TREE AND BRANCH COPPER TELEPHONE NETWORKS DSL Rings is a patent pending network architecture developed by Genesis Technical Systems Corp. The architecture provides most opportunity in the current tree and branch network if the portion of the network from the Central Office (CO), or Exchange, to the last Pedestal (Distribution Point DP) is bonded. Refer to Figure 1 for a graphical description of the architecture. Note that the links between the houses are implemented via passive jumper wires that do not come back to the Convergence Node (CN). In this way, a single CN design can efficiently manage 2 16 houses in a given ring. Genesis suggests a maximum of 16 houses in the ring due to the delay introduced by transiting each node to get back to the CO; however RPR has an upper limit of 255 nodes in a ring. DSL RINGS Figure 1 Current vs. DSL Rings Architecture HGW HGW DP Distribution Point CN Bonded Interface Ring architecture 2 copper pairs Shorter distance between VDSL nodes 2 new nodes - CN = GTS Convergence Node - HGW = GTS Home Gateway HGW HGW 1 Bonded pairs are used to obtain maximum bandwidth from the CO to the pedestal (DP). The Convergence Node, which is environmentally hardened and powered via the copper wire from the CO, terminates the bonded signals and acts as the gateway node for the subscriber collector ring. DSLRWP Genesis Technical Systems 4

5 As each node in the ring is a full ADM, based on VDSL2, the DSL transmission bandwidth starts at zero again on each individual hop. In most cases the hops back to the pedestal and then to the neighbour s house are less than 250 meters (<750ft). VDSL2 bandwidth at this distance is about 200 Mb/s (depending on VDSL2 chipset manufacturer s specifications and the quality of the cable). Please refer to Figure 2 DSL Rate Curves for comparisons. Figure 2 DSL Rate Curves Source: Infineon: Future Proof Telecommunications Networks with VDSL2 by Stephan Wimoesterer, Product Marketing Manager, VDSL2; July With DSL Rings there are two paths into and out of each house, each with the potential of carrying up to 200 Mb/s. Therefore the bandwidth potential for this scenario is up to 400 Mb/s (200 Mb/s Eastbound and 200 Mb/s Westbound) depending on the number of bonded pairs and the actual distance from the DSLAM to the pedestal. The greater the number of subscribers on the ring, the greater the bandwidth pool available due to the greater number of pairs available for bonding from the pedestal to the CO. DSLRWP Genesis Technical Systems 5

6 Figure 3 Throughput vs. Distance provides a rate comparison of standard VDSL2 with DSL Rings using various numbers of subscribers Bonded DSL Rings Bandwidth vs. Distance from CO Standard VDSL2 Standard ADSL2+ BDR Bandwidth Distance (m) RPR PROTOCOL DSL Rings technology is based on the RPR protocol that provides multiple advantages. The technology enables a fail safe in that, if a single pair is cut, the traffic goes in the opposite direction around the ring to get to the network gateway node. This is extremely useful for maintenance purposes, as well as for adding and removing nodes (houses) to/from the ring. This allows for a deployment business case based on customer demand which eliminates the sunken investment in a build it and they will come type approach. RPR also provides built in Quality of Service (QoS) for traffic differentiation and managed services as well as an Efficient Multicast (EM) capability that significantly reduces overall ring bandwidth requirements for multicast/broadcast video. RURAL AREA COVERAGE Within the DSLR architecture the bonded link to the CO/Exchange, which is typically a binder group (20 25 pairs depending on the telco), is terminated at the pedestal where a ring is initiated. In rural and suburban areas the pedestals are often connected together by the same DSLRWP Genesis Technical Systems 6

7 physical binder cable. The cable comes out, drops a few pairs to service a few homes, progresses down the street, drops a few more pairs, etc. The pairs that were dropped at the first pedestal are still physically in the cable bundle that progresses down the street to the next pedestal. DSL Rings provides the capability to, not only terminate the bonded link from the CO but to also, initiate another bonded link towards another pedestal down the road. It is thus possible for DSLR to provide up to 400 Mb/s bandwidth to homes that are greater than 7 km from the CO/Exchange, using the existing copper wire infrastructure, depending on the distances from the CO to the first pedestal and between pedestals. Figure 3: Typical Rural Telephony Deployment DSLRWP Genesis Technical Systems 7

8 Figure 4: DSLR Rural Implementation DSLRWP Genesis Technical Systems 8

9 NETWORK EVOLUTION Figure 5 depicts an evolution of the current network based on an expansion of the metro architecture into the last mile using hierarchical optical RPR rings deployed to the pedestal with the Genesis DSL Rings as the last 100m access technology. Figure 5 Evolution of the Network Central Office Central Office Optical RPR Ring Central Office Central Office Convergence Node DSLR Ring Optical RPR Ring Convergence Node DSLR Ring Convergence Node DSLR Ring Convergence Node DSLR Ring Convergence Node DSLR Ring The above architecture eliminates layers of transmission equipment (e.g.: DSLAMs) and the need for DSL Bonding thereby standardising the overall network topology and technology. Layer 2 RPR extended to the CPE provides a clean, efficient and fault tolerant packet handoff between hierarchical ring layers. RPR also simplifies device management, supports QoS and EM capability, DSLRWP Genesis Technical Systems 9

10 which can be used for live broadcast events and CPE software upgrades. The copper telephone line can be used to supply electrical power to the convergence nodes and CPE devices. DSLAM BYPASS The DSLR bonded backhaul link can be logically considered to be a single communications channel of relatively significant bandwidth. Accordingly DSLR blades could be produced that would fit directly into Edge Routers, Broadband Remote Access Servers (BRAS s) or Multi Service Access Nodes (MSANs) thereby bypassing the Central Office DSLAM altogether. DSLRWP Genesis Technical Systems 10

11 IV. Discussion A. Model Design 1. What wireline network technology and design should the model use to calculate costs, and how should the model calculate the terminal value of the network? Looking at work done in the UK as a reference and realizing that there are much greater distances and many more people in America even if the total proposed $9 billion over 5 years was matched by participating Telcos, it would not put much of a scratch in the overall need in America if optical fiber- based approaches were to be considered at all. In the UK the Broadband Stakeholders Group commissioned a report by Analysys Mason (BSG Report: The costs of deploying fibre- based next generation broadband infrastructure 8 September 2008) on installing optical fiber to the whole of the UK. Their report showed a cost of 24.5 billion to install FTTP (GPON) nationwide, 28.8 billion to install FTTP (Point- to- Point, e.g.: Gigabit Ethernet), and 5.1 billion to install FTTN/VDSL. (The UK (British Isles) cover an area about the size of New Mexico.) xdsl- based approaches make use of what has been called the trillion- dollar asset of incumbent US telcos the copper twisted pairs that go into each residence. Fiber- based alternatives look to replace this still useful asset at enormous expense. The architecture of the telephone access network has not changed in the 130+ years since the telephone was patented. By making intelligent technology choices it is possible to achieve huge increases, potentially in the vicinity of 100 Mb/s, in available bandwidth at distances even beyond 23kft. This seems incredible as the standard DSL transmission curves fall off quickly with distance: 1

12 For example, rural & hamlet telephone system infrastructure often looks something like the following: The message here is that, to achieve maximum DSL- based bandwidth, the FULL CABLE (24- pairs is often called a binder group in telecom speak) needs to be terminated at the first pedestal in the diagram above. If the bandwidth at that distance is only 4 Mb/s/pair the technology exists (G.Bond for DSL- layer or Ethernet Link Aggregation at the TCP layer) to aggregate the available bandwidth/pair into a single communications link. The result in the picture above would be 4 Mb/s x 24 pairs = 96 Mb/s that can be made available to all downstream customers. As can be seen, the distances after that first pedestal are not insignificant either. The obvious solution is to re- bond all pairs going downstream to the next pedestal in the same manner that they were done from the Central Office in the first place. In these pictures it is often easy to forget the physical aspects of cabling at the first pedestal (or cabinet) the twisted pairs that serve the nearest customers are not physically removed from the cable (binder). If the cable was installed correctly those pairs were simply cut at the punch block where they connect to the drop wires that travel the last 100 yards or so to the house. If they weren t cut it is called a bridged tap and is a real problem for xdsl- based systems. The point is that the physical pairs are still in the cable that goes down the road to the next pedestal. As can be seen in the picture above, it is possible to re- use those pairs to help in providing service to those customers who are downstream. If the distances 2

13 happened to be as in the picture, all customers would have access to 96 Mb/s at all times. In this case, those beyond the first pedestal would not have ANY access to xdsl- based Internet services. The bandwidth would be shared. As bandwidth in the telecom network is ALWAYS shared; it is just a question of at which point the sharing begins. Most people would probably say that being able to share a large bandwidth is better than having none at all. As this bandwidth has been achieved over the EXISTING copper infrastructure the costs are a fraction of the cost of optical fiber- based alternatives and the deployment times are measured in days as opposed to months. This system requires some intelligence in the pedestals and in the home modems that is not there today. It also requires mains power at the pedestals. To minimize delays in traffic that is sensitive to delays (e.g.: voice, video conferencing, etc.) it is necessary to have the system add/drop the traffic from upstream/downstream in the pedestals. To be cost efficient, the system needs to have a single piece of equipment that applies to however many of houses are served by each pedestal with a minimum of unused resources (e.g.: ports). Unused ports represent significant amounts of stranded capital investment for the Telcos. They are also a continuous electrical power drain for no benefit to the telco or customer. The core optical network between Central Offices (COs) solves this by using optical rings. This architecture also adds resiliency in the case of fiber cuts. Applying these same concepts to the access network is a technology called DSL Rings (DSLR) that has been shown to work and is being developed by Genesis Technical Systems of Calgary, Canada & Coventry, UK. The resultant network looks like the following: 3

14 This architecture can solve the broadband divide in almost the entire US while also providing the deployment platform for next generation wireless services provided by femtocells. The question then becomes: if spending $9 billion can deliver ultra high bandwidths such as those described above over completely brown- field copper infrastructure for likely 17 million of the 18 million unserved, why consider spending the same amount and delivering similar bandwidths over fiber- based infrastructure to probably much less than 1 million? 4

15 Technology Type Fiber-ONLY Copper-ONLY - NOTE: Copper- ONLY can also be used in the Fiber/Copper hybrid case Fiber/Copper hybrid FTTP/H (Fiber To The Premises/Home) DSL Rings - Genesis Technical Systems FTTN (Fiber To The Node) FTTC (Fiber To The Curb) Comparison Description xpon (Passive Optical Network) A single optical fiber runs all the way from the CO to the neighbourhood where it 'passes' each house in that neighbourhood. When the customer requests service a technician comes out to 'connect' the house to the single fiber strand via an optical splitter or coupler. In some cases Verizon removed the copper wires from houses where they provided xpon services so that the customer had no fall-back option. Optical Ethernet (often called Point-to-Point) A 'cable' of fibers is routed from the CO to the neighbourhood. There are sufficient individual fibers in the 'cable' (or 'bundle') so that there can be a single fiber routed to each house in that neighbourhood if every house requests service. When service is requested by the customer a technician comes out to physically route their individual fiber from the 'cable' to the customer's house. This is similar to the current copper network architecture except using fiber. Economic Criteria The 24-to-50-pair cables from the Pedestal to the CO are logically bonded together to create a single communications link. This can be done as a Plain Old Telephone Service (POTS) overlay so that Emergency voice services are not impacted when the power fails. From the pedestal to the home a copper ring is created by connecting home-to-home in the pedestal using a passive cross-connect matrix (which enables adding & removing houses from the ring remotely) and using 2-pairs/home. Home Gateway devices in each home terminate (traffic can be added, removed, passed through, or drop-andcontinue multicast) and regenerate the signal on the existing copper lines so that the VDSL2 transmission curve re-starts at each home. This enables very high bandwidth over the existing copper telephone wires. Fiber is routed from the CO to the big cabinet (typically more than 6 feet across and 4 feet high) - often green in colour. A Remote DSLAM is installed in this cabinet along with the big wiring patch panel. The fiber link basically replaces the big pair cable coming from the CO thereby reducing the distance that the DSLAM has to transmit. This makes the resultant DSL-based bandwidth higher as the transmission distance has been reduced by the distance from the Cabinet to the CO. The link from the cabinet to the house is still the existing copper plant. Often VDSL2 is the DSL technology of choice from the Cabinet to the home. Fiber is routed from the CO to the Pedestal. This bypasses the pair copper cables from the CO to the Cabinet and the 24-to-50-pair cables from the Cabinet to the Pedestal. A mini-remote DSLAM is installed in the pedestal. The original pedestal - basically a small wiring patch panel - is replaced with a new enclosure that has mains power and batteries. This is an outside plant, environmentally hardened enclosure. Cost Efficiency How well does the capital expenditure scale with adding/removing customers? What is the potential for stranded capital investment? Highly inflexible physical architecture. Fiber is deployed in the hopes that customers will buy the services offered over that fiber. A neighbourhood has a single fiber deployed to it, then a truck roll has to occur for each customer. High potential for stranded capital investment. Highly inflexible physical architecture. Fiber is deployed in the hopes that customers will buy the services offered over that fiber. A neighbourhood has the fiber 'cable' deployed to it, then a truck roll has to occur for each customer. Highest potential for stranded capital investment. Once installed there are no physical changes that need to occur to add or remove served customers. The potential for stranded capital investment is comparitively nonexistent. Minimal customer impact other than seeing higher bandwidth over their DSL connection. May need a new modem to achieve the higher rates. However, least return for Telcos in that all it offers is increased bandwidth of 2-4x non-fttn systems Not as cost efficient as FTTN. High initial costs to supply a neighbourhood with the risk of low take rate on premium services. Higher bandwidth is the only advantage of this approach over straight DSL or FTTN. Deployment Time Once the decision to deploy is made, what are the various logistical obstacles to providing service to a selected area Very long as new fiber cable must be trenched or hung from poles, legal Rights-of-Way must be secured and batteries installed in homes for cases where the power fails. Very long as new fiber cable must be trenched or hung from poles, legal Rights-of-Way must be secured and batteries installed in homes for cases where the power fails. Very fast - it is estimated that a single technician can deploy DSLR to 4-5 houses, including the pedestal upgrade, in a single day. Faster deployment than FTTP/H but still requires significant amounts of fiber to be deployed as well as mains power to the cabinet, a new cabinet with batteries & Heating, Ventillation, & Air Conditioning (HVAC) systems as well as the remote DSLAM. Much longer than FTTN but not as long as FTTP/H. Needs mains power to terminate the optical signal at the pedestal which generally implies batteries and HVAC systems as well. Deployment Model Does the system have to be deployed before the customer can be served or can the customer be added to the service later on in an efficient manner? Describes the level of risk to deployment capital used. Build it and they will come - highest level of capital risk in telecom networks today. Build it and they will come - highest level of capital risk in telecom networks today. Pay as you grow - lowest level of capital risk in this spreadsheet. Capital is not risked until the customer agrees to pay for the service. Build it and they will come - Third highest level of capital risk in this spreadsheet. Build it and they will come - Second highest level of capital risk in this spreadsheet. Technical Criteria

16 Bandwidth What are the announced rates of traffic delivery in Megabits per second (Mb/s) Set bandwidth slice. Current advertised options include: 5Mb/s, 15Mb/s, 30Mb/s, and up to 100Mb/s in some cases though this is configurable by the telco up to the split ratio of generally 1:32 (i.e.: one fiber is shared with 32 homes) Each house gets a dedicated optical fiber so whatever the end points can support is what can theoretically be offered. 1Gb/s is being offered over this type of architecture in some areas of some Asian countries (e.g.: South Korea, Tiawan, Japan, Singapore). Up to 400 Mb/s combined upstream & downstream (2x VDSL2 maximum bandwidth) depending on the number of upstream pairs bonded and distance the pedestal is from the CO. Typical maximum of 20 Mb/s unless 2-pair bonding from the cabinet is used, which would provide up to 40 Mb/s depending on the distance between the cabinet and the home and the DSL bandplan used. Typical maximum of 80 Mb/s for a single pair or 150+ Mb/s for 2-pair bonded. Depending on frequency band plan, 200 Mb/s symmetric (400 Mb/s combined) could be achieved with VDSL2's 30a bandplan. Power How is power provided to the system? Downstream power supplied by CO, upstream power supplied by consumer mains. Downstream power supplied by CO, upstream power supplied by consumer mains. All power to the pedestal is supplied by the CO in urban situations. Mains power required for rural applications. Lifeline capability provided by POTS support. Mains power has to be supplied to the cabinet, which generally has only a passive wiring patch panel in it today. The household modem is powered by the household mains. Mains power required to terminate the optical signal which also suggests batteries and a larger enclosure than the standard pedestal. Lifeline Support Sharing Security Quality of Service Multicast Support Infrastructure Re-Use Scalability How does the system provide lifeline support when the power fails? What if a natural disaster (Katrina, ice storm) knocks out power for an extended period where mobiles will not be able to be charged from the home? All network bandwidth is shared at some point. What is the mechanism for sharing the bandwidth? What are the traffic security considerations in each system? Given that all bandwidth is shared, are there any provisions for prioritizing different kinds of traffic at higher levels than others? Are there any built-in provisions for supporting large broadcast events? Deployment timeframes and community disruption are generally lower if the existing infrastructure is being re-used How does the system scale with the addition and removal of users? Lifeline capability provided by batteries in CPE. Typical battery life is 8 hours or less. PON technology broadcasts all downstream data to all houses so all downstream bandwidth is shared. Upstream bandwidth is dynamically assigned timeslots based on various parameters such as transmit queue fill levels. Encryption is used to protect each user's data. Otherwise it is a bus model and everyone can 'see' everyone else's encrypted traffic. Lifeline capability provided by batteries in CPE. Typical battery life is 8 hours or less. Sharing happens once the traffic reaches the CO in the larger network 'cloud'. Most physically secure as the fibers do not go near the other customers and are not shared. Can be implemented in the upstream (customer to network) direction but Not really necessary until the traffic downstream priorities are set by the hits the wider network 'cloud' network. PONs are broadcast media so this can be implemented easily. Depends on the capabilities of the router that terminates the fibers going to the individual houses. This has little impact on the bandwidth seen by the individual customers. DSL Rings can be implemented as a frequency overlay on top of POTS. POTS has been shown to survive weeks without power as long as the CO power is maintained. Each DSL Ring-based Home Gateway is an Add Drop Multiplexer (ADM). All ring bandwidth is available at each node subject to QoS and SLA provisions. All traffic is encrypted and the Convergence Node in the pedestal is firewalled. There is also an aspect of physical security as a user's data only goes in one direction around the ring. The closer the house is to the pedestal the more traffic that can be seen assuming the technology and understanding. Can be supported around the DSL Ring to the pedestal or throughout the network as a whole. Service Level Agreements can be offered based on traffic priorities so that the network becomes a revenuegenerator again. Can be supported around the DSL Ring to the Convergence Node in the pedestal or throughout the network as a whole. Generally no battery back-up for power failure situations. The copper portion of the link between the cabinet & the home is not shared but the fiber link from the cabinet to the CO is shared between however many houses are served by the cabinet (typical numbers would be in the several hundreds). Physical security is strong (as the wires do not pass through neighbour's houses). Encryption is generally not used because of this. Most telcos do not implement QoS in these networks as the link to the cabinet is not shared. Unfortunately this means that all traffic is treated the same once it hits the DSLAM. Depends on the capabilities of the Remote DSLAM that sits in the cabinet. This has little impact on the bandwidth seen by the customer over the copper in this architecture. Generally no battery back-up for power failure situations. The copper portion of the link between the pedestal & the home is not shared but the fiber link from the pedestal to the CO is shared between however many houses are served by the pedestal (typical numbers would be in the 3-16 range). Physical security is strong (as the wires do not pass through neighbour's houses). Encryption is generally not used because of this. Most telcos do not implement QoS in these networks as the link to the pedestal is not shared. Unfortunately this means that all traffic is treated the same once it hits the Remote mini-dslam. Depends on the capabilities of the Remote DSLAM that sits in the pedestal. This has little impact on the bandwidth seen by the customer over the copper in this architecture. None None Complete Partial Partial Most PONs are split at less than 1:32 but the split ratios can be up to 1:128 for GPON or 1:32k for EPON. The base rate is Gb/s. Basically, PONs subdivide the available bandwidth between the customers who are using the fiber. Logistical Criteria The fibers are there and 200+ fiber 'cables' don't cost a lot more than 10- fiber 'cables'. If the number of fibers needed exceeds the fibers in the 'cable', a whole new cable must be laid. Customers may be added or removed at will while maintaining service to all existing customers. Whatever bandwidth is available over the fiber is split amongst the several hundred homes served by that cabinet. Depends on the granularity of the Remote mini-dslam that resides in the pedestal. If there are more customers than ports provided by the mini-dslam, another box will need to be added. Additional fiber(s) may or may not be required depending on whether Wavelength Division Multiplexing (WDM) is used on the backhaul fiber.

17 Network Applicability Stranded Capital Investment Managed Services Rural Application ROI Timeframe Risk Profile Can the system be economically deployed on a National basis? What is the relative potential for capital to be spent and not utilized for revenue generation purposes? How is the Telco able to provide services that they can charge a premium for? Is there an economical case to be made for deployment in rural areas? How long will it take for the upgraded network infrastructure to generate an operating profit? [estimated] What is the risk that the capital spent will not generate a profit for the Telco? Not economically justifiable to more than an estimated 20% of homes in the USA. Not economically justifiable to more than an estimated 20% of homes in the USA. Complete Not likely to be economically justifiable to greater than 60% of a Nation's population depending on demographics and ROI targets. Not likely to be economically justifiable to greater than 40% of a Nation's population depending on demographics and ROI targets. $$$$$ $$$$$+ $ $$$ $$$$ Carrier has complete control None Conclusions Only matters in the network 'cloud' as there is really no restriction on the potential bandwidth available to the home. None Carrier has complete control Definite application, ROI Timeframe may be somewhat higher than urban application and would benefit from government rural incentive programs Only matters from the network to the cabinet as the customer sees a dedicated link. Minimal Only matters from the network to the cabinet as the customer sees a dedicated link. >10 years >10 years <2 years >5 years >8 years Highest Highest+ Lowest by far Middle High None DSL Rings is Better Because DSLR is generally less than 1/10 the cost to deploy in urban areas and 1/100th the cost in rural areas, can be deployed much faster, gives Telcos more time to decide what fiber-based option they really need, can provide POTS when the power fails, etc DSLR is generally less than 1/10 the cost to deploy in urban areas and 1/100th the cost in rural areas, can be deployed much faster, gives Telcos more time to decide what fiber-based option they really need, can provide POTS when the power fails, etc N/A DSLR can be deployed in DSLR can be deployed in conjunction with FTTN making the conjunction with FTTC making the FTTN deployment look even better FTTC deployment look even better to the consumer. Instead of FTTN, to the consumer. Instead of FTTC, DSLR can be deployed more quickly DSLR can be deployed more quickly and economically in far more and economically in far more scenarios/situations. scenarios/situations.