Today, the world’s telecommunications companies are confronted with perhaps the biggest challenge they have ever faced: how to migrate to an all-IP, next-generation network (NGN). It is a challenge they must handle successfully because already, ARPU (average revenue per user) is falling.
Why is this happening? Principally, because broadband has become part of everyday life. Users expect at least 0.5 Mbit/s, maybe 8 Mbit/s and in some countries they already enjoy 100 Mbit/s into their homes. xDSL broadband has become a standard service, and more and more bandwidth is becoming necessary in the core and access networks to satisfy demand.
At the same time voice services have ceased to be profitable for many network providers due to deregulation, competition, and because many telcos own legacy networks using multiple technologies, which are no longer efficient enough to provide the low operating cost-base essential for profitability.
An ubiquitous bit-transport system for all network services — an all-IP NGN — is a way to drive significant operating costs out of the network. Fundamental changes in the service they offer, together with increasing competition, mean operators must migrate to an all-IP NGN because of the huge efficiency, capacity and flexibility it will provide.
In addition to driving out operating cost, network operators see the provision of “content” across their packet delivery networks as their financial salvation. Video on Demand is probably highest on this list, typically as part of a Triple Play (voice, data, video) offering to their customers. However, both the consumer demand for faster downloads and the operators' desire to sell high-definition (HD) video-on-demand have requirements that the existing network infrastructure just cannot satisfy.
While a “poor man’s” Triple Play can be offered on ADSL2+, with a minimum of 10 Mbit/s, this technology is not suitable for HD television/video. And as the maximum 22 Mbit/s figure can only be achieved within 1-2 km of the central office (CO), this excludes many customers. Ultimately, operators must provide true Triple Play, requiring up to 55 Mbit/s to each home.
Next stop: Multi-Service Access Nodes
Every incumbent operator’s network will, of course, be different, but as they mostly grew from a very similar telephony background they will in all probability use similar technologies albeit deployed to greater or lesser extents.
The core network will often be ATM/SDH/SONET with TDM digital switches at the CO and remote units of these switches in suburban and rural areas. The ATM core or trunk network is used to interconnect the Class 5 CO switches to higher order Class 4 (trunk) and international switches. This same ATM network is used to backhaul broadband services from DSLAMs located at the CO to ISPs and internet gateways (after conversion to IP/Ethernet).
Many large operators will start the NGN migration by concentrating on converting the core network to IP. Once this has been accomplished, all the existing CO switches can be retired and replaced by MSAN equipment — some at the CO buildings and some in outside plant (OSP) active equipment cabinets.
The ultimate broadband solution would be to provide fibre into every home and business, giving a bandwidth of several Gbit/s. However, many operators view an immediate move to FTTH/P (fibre-to-the-home/premises) as unfeasible — the investment required is vast — and want another 10 years use from the vast amount of copper in the ground.
Hence, operators worldwide are already deploying the MSAN strategy, or actively considering it as part of their network upgrades. In Europe, these include Belgium’s Belgacom, France Telecom, Telecom Italia, Swisscom, Telekom Austria, Deutsche Telekom, and Norway’s Telenor. One of the largest NGNs being built is BT’s £10 bn 21CN project, which makes extensive use of MSANs.
In the Far East, Israeli company ECI Telecom is expanding MSAN deployment in countries such as Taiwan, Vietnam and the Philippines. Argentinean telecoms operator Telpin is building IP-based networking services facilities using iAN-8000 MSANs from UTStarman. And by implementing Nokia Siemens Networks MSANs, STC Saudi Arabia increased its broadband penetration in 2008 from 4 to 22% in just nine months.
Making the switch
Ideally, from an operational perspective, all the MSAN equipment would be located in the CO building, and a similar program employed to that of 20 years ago. Then, TDM digital switches were built alongside their analogue predecessors and then a grand ceremony of “pulling the wedges” took place, with every customer on a CO being instantaneously cut-over onto the new switch.
Sadly, this is not feasible with next-generation network upgrades. Delivery speeds of ADSL2+, VDSL or VDSL2, are all severely limited by distance, typically to about 1 km. In many existing CO service areas however, the customer can be up to 5 km away — beyond the reach of receive high-speed services. Therefore, the MSAN equipment providing the broadband service has to be moved out of the CO building to a place less than 1 km away from the customer. This dictates that the MSAN has to be located in outside plant (OSP), logically at or alongside the existing cross-connect or distributor cabinet (often at the roadside and usually serving some 300 to 600 customers).
This approach requires a different network topology to that which currently exists. Current CO buildings containing Class 5 switches and DSLAMs will be replaced by MSAN equipment (equipped with PoTS, ISDN and DSLAM cards) and mostly housed in hundreds or thousands of OSP cabinets. All of these OSP-based MSANs will require IP-over-fibre connections to the core network — also known as fibre-to-the-node (FTTN).
The MSAN cards convert xDSL, ISDN, POTS and other services to IP and this is then backhauled into the core. Consequently fibre has to be dug in from the core point-of-presence (which may not be the existing parent CO building). In some cases, it may be possible and cheaper to pull-in fibre through existing ducts. The use of blown fibre to these MSAN locations should be seriously considered as this provides future-proofing, preparing for the ultimate conversion to FTTH/P in future.
Converting the core network to IP is a massive job which will take several years and over that period faster DSLAMs will no doubt be deployed, so operators must consider how such deployments will fit into the overall migration plan. For example, even if new DSLAMs are installed in the CO they should be IP-based, not ATM-based, using IP to ATM converters. Then, when the new IP core is ready, they can simply be re-parented (dispensing with the converters) rather than replaced.
Use MSAN not DSLAM
In fact, they should not be DSLAMs at all. A better strategy would be to deploy MSAN chassis equipped with DSLAM cards. Then, when the IP core is ready, POTS, ISDN and other cards can simply be added into these same chassis, again avoiding the need for total replacement. In some areas at least, operators will be providing VDSL/VDSL2 high-speed services, requiring the DSLAMs or OSP cabinets to be within 1km of the customer. These are the locations that will ultimately become MSAN nodes. Therefore by deploying MSAN chassis with DSLAM cards, in due course, all that is needed is to insert POTS, ISDN, etc., cards when the old CO switch is ready to be decommissioned.
MSANs can only provide their full services when parented on an all-IP core network with the necessary “softswitches” for call control and routing functions previously provided by the Class 5 switch. Thus, in the interim migratory steps just outlined, MSAN chassis and DSLAM cards are used only to provide DSLAM/broadband services. Other services like POTS and ISDN continue to be provided over copper from the CO to the MSAN cabinet, where splitters will combine the voice and broadband signals onto the copper pair to the customer. In all probability, the IP-over-fibre signals from the DSLAM-equivalent MSAN will be converted to ATM at the CO for onward transmission through the existing “ATM cloud” to internet gateways.
Once the IP core is complete, sufficient MSANs will be needed to provide services to all of an existing CO’s customers. At the OSP distributor, customer lines need to be transferred from the CO copper pairs onto MSAN ports. Where possible, customers should be persuaded to move to VoIP for their telephony since this means that voice, data and video can all be provided as Triple Play from a DSLAM card — achieving IP all the way to the customer. If it is not possible to move the customer to VoIP, they can be reconnected to the appropriate type of MSAN card port. Once all customers of a particular CO have been migrated, the old CO switch, ATM DSLAMs and other equipment can be decommissioned. When all existing CO switches have been retired, the network becomes all-IP to the MSAN nodes and in many cases all-IP to the customer.
The fibre between an MSAN and its parent core network point of presence (PoP) could be up to 20 km long, many times the distance between the cross-connect or distributor it replaces and its current parent CO. This means far fewer core PoPs are actually needed. So potentially thousands of CO buildings — often in prime locations — can be sold, helping to finance the NGN migration. Of course the migration plan needs to identify at an early stage where MSANs will be parented and which CO buildings can be sold.
Preparing for the future
As stated earlier, in the longer term operators are expected to move to FTTH/P. The strategy outlined can support this if operators run in blown-fibre ducts to the nodes. The next stage of migration would be to convert nodes from active MSANs to passive fibre distribution hubs (FDH). For each node’s service coverage area a fibre dig (or in some cases overhead distribution) would be needed to lay one or more fibres from each home or premises back to the node. There are two options for FTTH/P, a passive optical network (PON) using shared fibres, which are more cost-effective for long CO to node distances such as in rural locations, or a point-to-point (P2P) for metro and urban situations, where every customer has their own fibre back to the IP-core. Most operators will need a mix of both.
For PON, the MSAN will be replaced with a fibre distribution hub where passive optical splitters combine the lightwave signals from 32, 64 or 128 customers on to a fibre pair back to the IP-core. Alongside this an optical fibre cross-connection or patching frame (similar to that we currently have for copper) to provide moves, adds and changes (MAC) flexibility. For P2P, a lot of new fibres back to the IP-core need to be blown-in or dug in. In this instance, the fibre distribution hub only needs to provide the fibre cross-connection facility.
In both cases, the node or fibre distribution hub reverts to being completely passive — reducing field maintenance and associated costs significantly. And, as previously, this conversion to FTTH/P can be undertaken progressively — either a node at a time or even by placing fibre distribution hub and MSAN side-by-side in the OSP and performing a gradual change over as customers are prepared to pay for FTTH/B enabled services.