Written on 3:52 PM by ooe
To date the use of optical fibre in the access network has typically only proved cost effective for supplying the high bandwidth demands of large corporate customers. For new and established network operators, however, the increasing demand for bandwidth to deliver bearer, interactive and bundled services to business and residential customers is requiring them to seriously consider the high volume rollout of optical fibre based systems. Network operators therefore face some major decisions, not only in terms of the type of fibre transmission systems to deploy, but also how to install a cost effective network of fibre cables, ducting and joints to connect to the customers they wish to serve. Installing fibre in the ground represents a major commitment and a long term investment. Network operators can typically expect the fibre infrastructure to equate to at least 60% of the cost of the overall access transmission system. Planning and building such networks is a major upfront investment, and in today's highly competitive markets operators are faced with the added complications of uncertain take up of services by customers and the likelihood of high customer churn.
This chapter focuses on fibre to the home (FTTH) and the deployment options and challenges for the physical fibre infrastructure. The key difference associated with connecting optical fibre to residential properties instead of business properties is not technical but commercial. There is typically a much smaller potential revenue from a residential property than from a business property. This leads to a need for cost optimisation of both the transmission system and the fibre infrastructure. The chapter also examines the fibre infrastructure issues that need to be considered for a FTTH deployment and describes the fibre access network modelling activities carried out by BTExact's Broadband Network Engineering Unit. The aim of this modelling is to automatically plan and cost optimise the deployment of access networks based on real geographical and demographic data.
BT's telecommunications network in the UK consists of two main segments. First, a core network of approximately 5500 exchanges (switching sites) connected by high speed, mostly optical fibre, links. Second, an access network connecting customers to the switching sites. For large, and increasingly for medium sized customers, this access network is also based on optical fibre. The fibre access network is currently carrying traffic at anything from 2 Mbit/s to hundreds of Mbit/s. For residential customers the access network consists of twisted copper pairs contained within a tree and branch cable network. This network design has not changed appreciably for 50 years or more. Indeed some cables have been in service for that long.
|
Written on 3:51 PM by ooe
A constant theme through this chapter is that technology is improving all the time and nowhere more so than in the access network space. It is clear that point-to-point technologies will improve with time, however, they tend to require concentrators in the access domain or long runs of optical fibre, both of which are non-optimal except for green field sites. Despite these limitations, many manufacturers and start-ups are active in this space and future upgrades to this type of equipment are certain.
Conclusions
The pace of change in the electronics world has never been greater and this is leading to changes in PON technology which will impact significantly on the price/performance characteristics of new fibre access systems. Furthermore, there are many new companies trying to create systems that will become successful in the access network area.
Access network bit rates are increasing from the original 155 Mbit/s (20 Mbit/s for non-standard systems) to 1.25 Gbit/s and seem set to continue to rise. Whilst these bandwidth increases might not be useful in all applications, there is potential for these systems to stray into areas previously the domain of core technologies such as SDH or SONET. It might be that the introduction of new PON technologies results in a reduction in price of the older fibre access systems – making them more cost-effective in the access network space.
Whilst telecommunications operators are presently rolling out ADSL technologies and will be doing so for some time, there is already clearly a case for higher speed access network technologies being introduced into the networks of the world in the near future. Whilst VDSL may have a place in this market, the optical fibre technology must provide the ultimate solution given that the PON was specifically designed for this market segment.
Whilst point-to-point systems will have a part to play, their use as a ubiquitous access network architecture seems unlikely, given the advancements being made in PON technology.
Of course there are questions as to when the cost-reductions and price breaks that always follow new technology will occur, but this cost reduction will happen. It seems that it is not ‘if’, but merely ‘when’ fibre technologies are adopted in the access network as a matter of course, with early adopters perhaps getting the largest customer base. It must not be overlooked that laying optical fibre into an access area is not cheap, and that the economics for a second operator are so much poorer and so a second optical fibre access network in an area will probably never happen.
|
Written on 3:51 PM by ooe
Current PON technology is limited to 622 Mbit/s, however, this is not a practical limit for this technology, merely the highest agreed bit rate supported by the ITU-T. The IEEE and FSAN are already considering systems with 1.25 Gbit/s PON speeds, and higher bit rates are feasible.
Whilst the IEEE Ethernet PONs are rumoured to be a lot cheaper that APON technology, so far no systems have been produced to test this hypotheses. Time will tell whether this technology will actually prove to be cheaper although many components will be common between the two architectures leading to some doubt as to the reduced cost claims.
Already we have seen experimental Super-PON architectures from Alcatel [3,4] where the split ratios can be increased from the more usual 32 to over 100, with thousands or more being possible. Furthermore, optical amplifier technology is making possible PONs with a transmission reach of thousands of kilometres.
WDM technology will allow common infrastructure to serve both residential and business end users, even if those businesses require very high point-to-point data rates.
|
Written on 3:50 PM by ooe
Point-to-point systems generally utilise multiplexors at or near the street cabinet. This is not ideal for most ILEC applications although CLECs may find the technology useful in the longer term as this fits with their current architectures.
It is fair to say that point-to-point architectures usually compare favourably to PON architectures where there is a strong demand and a central site at which concentration can be achieved, and hence they are often used for campus environments.
Point-to-point technology has been used successfully where there is a need for fast Internet service, however, the systems tend to be poor at providing voice services. For that reason they seem best suited as an overlay technology, leaving existing services on copper. For the network operator, however, the costs of maintaining two parallel networks can prove highly undesirable.
|
Written on 3:50 PM by ooe
Most operators are rolling out ADSL systems and are considering the next step ahead. VDSL is considered a realistic future step for the access network. However, all DSL technologies have one major weakness, they require a metro-type transmission network to feed them. Some in the industry are experimenting with new feeder architectures based on PON designs. Current 155 or even 622 Mbit/s systems are not really powerful enough for these core feeder applications but future PONs based on 1.25 Gbit/s and higher seem to be well suited to these applications.
One could visualise an architecture where the access and metro/core networks could be rolled into one and served by powerful PON systems. Studies are already underway to examine the cost implications of such architectures and if successful, these could pave the way for the introduction of the technology in a much shorter timescale than would otherwise be possible. These systems would need greater split ratios and much longer reach than current PON technology would allow.
|
Written on 3:50 PM by ooe
As has been said many times, optical fibre in the access network would create the ultimate access network, although controlling the costs of implementing such architectures is key. PON technology has been shown to reduce the costs of access fibre network rollout for two reasons:
-
It allows a reduction in the number of fibres back to the exchange giving better duct utilisation and shared head-end (OLT) costs.
-
It inherently concentrates the traffic onto fewer data streams hence reducing the cost of metro equipment.
Given the increase in speed of computing devices, and the proliferation of the Internet, it is to be expected that the market for high-speed connections will continue to grow and will become ever more resource hungry, both in terms of bandwidth connectivity and service support. In this market it is to be expected that optical technology will outstrip other access network technologies to become the technology of choice.
|
Written on 3:49 PM by ooe
PON systems have never be deployed in quantity in Europe for several reasons. Whilst it is true that these systems need to be multi-service, it is also fair to say that most systems to date have not proven to be so with PSTN services missing from most designs. This situation is now changing and several manufacturers are showing PSTN on their PON product roadmaps, with delivery of such systems scheduled for mid 2003.
Despite the costs and lack of flexibility associated with current equipment designs, there are inherent advantages of PON technology and the next sections look at types of deployments for which PON systems are most suited.
|
Written on 3:21 PM by ooe
In order to create a complete system specification for telecommunications equipment, it is necessary to detail the managed entities and their scope as well as degree of interaction. For the APON specifications, OAM (operations, administration and maintenance) functionality is specified in two areas. First, we have Recommendation G.983.2 (‘ONT management and control interface specification for ATM PON’), and second, we have G.983.7 (‘ONT management and control interface specification for dynamic bandwidth assignment (DBA) B-PON system’). These are complementary specifications.
The ATM-PON element management system will only manage ONTs as part of the ATM-PON system through the OLT using the ONT management and control interface.
The OAM requirements present requirements for the ONT management and control interface. First, it specifies managed entities of a protocol-independent management information base (MIB) that models the exchange of information between the OLT and ONT. Second, it covers the ONT management and control channel, protocol, DBA protocol, and detailed messages. Refer to the ITU-T specifications for more detailed information. For point-to-point systems, the situation is not so far advanced and no common specifications exist for OAM applications. 9.7
|
Written on 3:21 PM by ooe
It has been identified that PON systems might not give the level of reliability required for certain applications. Whilst the PON electro-optics are generally reliable, it has been found that a major cause of failure is the physical media, i.e. the optical fibres in the access network themselves, and in many cases faults are caused by external influences. Point-to-point systems can allow for protection using back up or standby fibre pairs.
It is not the aim of protection scenarios to restrict the architectures possible in a PON architecture but to support and enhance whatever architecture is chosen for a particular application. For this reason, the protection scenarios offered are not exhaustive, but are representative of those it is considered will be of interest. These must be carefully tailored to the applications for which they are required.
Generally it is only FSAN, and through them, the ITU-T that have considered protection of PON systems in some detail and an ITU-T recommendation is due shortly. This will be known as ‘G.983.5: A broadband optical access system with enhanced survivability’.
The main requirements of a protection system shall be that:
-
It should be possible to have several C protection configurations on the same OLT.
-
It should be possible to duplicate the OLT-PON-interface and the fibres between the OLT and splitter and to duplicate the entire fibre path between the OLT and ONU for a set of ONUs.
-
It should be possible to have a mixture of protected and unprotected ONUs on one B-PON interface.
-
The addition or removal of a protected ONU on a PON should not affect other ONUs on the same PON.
-
It should be possible to have automatic switching, which would be triggered by a fault detection such as loss of signal, loss of cell delineation, signal degrade (e.g., BER becomes worse than the pre-determined threshold), etc.
-
It should be possible to have forced switching, which would be activated by administrative events such as fibre re-routing, fibre replacement, etc.
-
It is necessary to avoid unnecessary switching. Because unstable switching affects service quality, unnecessary protection switching and unnecessary revertive protection switching should not occur.
-
It should be possible to realise switching without connection loss of the ATM connections.
-
It should be possible for the operator to choose between a revertive and a nonrevertive switching mode.
-
The service halt time should be less than 50 ms if the extra traffic option is not used.
-
The events or conditions that trigger automatic switching should be chosen among the G.983.1 OAM parameters.
-
The chosen protocols and mechanisms must apply to the B-PON section layer.
-
Extra traffic should be carried over the protection entities while the working entity is active and would not be protected. This capability will provide effective usage of bandwidth on the protection entities.
-
It must be possible for an operator not to activate the extra traffic option (e.g., to achieve a lower service interruption time).
Extra messages are introduced into the base set of PLOAM messages from G.983.1 to cope with the additional functionality required for protection switching but generally it is the OLT that initiates the switch-over.
There are two basic protection schemes that can be used and the authors do not attempt to judge failure mechanisms within this chapter and make no recommendations as to the better protection mechanism to use. The first is a partial protection scheme and is designated type B protection. It is useful when it is the feeder fibre(s) that are considered at most risk of failure. The more fully protected scheme is known as type C protection.
|
Written on 3:21 PM by ooe
Most access network systems, with the exception of point-to-point systems, use fixed, or static, bandwidth assignment. Most point-to-point systems allow static multiplexing and therefore the following section does not directly apply to them. With PON access network Systems, the situation is somewhat different.
A typical APON system would run at a bit rate of 155 Mbit/s, allowing a total customer bandwidth of around 130 Mbit/s. If this were statically multiplexed, the bandwidth per customer on a 32 way PON would be in the order of 5 Mbit/s. Now 5 Mbit/s is not a lot and could be provided by alternative technologies such as DSL.
Dynamic bandwidth assignment relies on the fact that it is uncommon for each user on a PON to require their full bandwidth at any one time. In fact each user can have potential access to bandwidths in excess of their mean bandwidths for extended periods of time and this is what dynamic bandwidth assignments would allow.
Instead of bandwidth being assigned on a per user basis, bandwidth is provided on demand, with the user unaware during normal operation that their share of the overall bandwidth ‘pot’ has increased.
On an APON system, it is not too difficult to arrange this as the speed of transmission over the PON is quite fast and therefore latency issues are kept to a minimum. The user is unaware of the precise mechanism by which the ATM packets are queued and then transmitted across the PON.
The principle recommendation from the ITU-T that deals with dynamic bandwidth assignment is G.983.4 (‘A broadband optical access system with increased service capability using dynamic bandwidth assignment’). Dynamic bandwidth assignment is also known as DBA.
In the DBA recommendation, G.983.1 is modified to allow for extra messages required in order to support the DBA mechanism. The DBA standard is designed to include:
-
Performance objectives (for example, bandwidth assignment delay, maximum waiting time).
-
Application functionality (for example, dynamic bandwidth assignment for bursty traffic and for ONU/ONT aggregated traffic composed of different traffic classes).
-
Fairness criteria and protocols (for example, dynamic bandwidth assignment based on ONU/ONT status reporting, dynamic bandwidth assignment based on OLT monitoring, dynamic bandwidth assignment based on a combination of reporting and monitoring).
-
Backwards compatibility.
As a background to understanding the dynamic bandwidth principles, it is essential to understand the way an APON works, at the cell level.
The basic APON standard is the ITU-T Recommendation G.983.1 which specifies a flexible access platform to provide broadband services through passive optical networks. In this Recommendation, the upstream traffic from the ONUs/ONTs to the OLT is transferred in a frame of 53 time-slots (cell slots). Each time-slot consists of three bytes of PON layer overhead and an ATM cell or a PLOAM cell. The upstream bandwidth is shared among the associated ONUs/ONTs. The OLT controls each upstream transmission from the ONUs/ONTs on a time-slot-by-timeslot basis. This is accomplished by sending data grants in downstream PLOAM cells.
PLOAM data grant cells are sent in the downstream direction to all ONUs/ONTs. The data grants are addressed to specific ONUs/ONTs and contain parameters that include the number of upstream data grants and the time-slots for the grants that are assigned to the individual ONU/ONT. Originally the grants were assigned in a static manner and only updated when a new connection is provisioned on a PON or an existing connection is removed from a PON. Once the bandwidth is provisioned, the OLT would continuously send the assigned grants to the associated ONUs/ONTs; subsequently, the OLT would receive corresponding user cells in the upstream. This current granting mechanism was efficient for real-time traffic but lacked any dynamic assignment capability.
However, the above mechanism is not efficient for non-real-time traffic, or where bandwidths are increased beyond the norm. For non-real-time traffic types, the ability to assign bandwidth dynamically is expected to provide higher efficiency than the original static granting mechanisms.
In ITU-T G.983.1, the OLT grant generation and distribution was updated when a new connection is provisioned on a PON or an existing connection was removed from a PON. Once the bandwidth is provisioned, the OLT would continuously send the assigned grants to the associated ONUs subsequently, the OLT would receive corresponding user cells in the upstream.
However, ITU-T G.983.1 was intended to enable a wide range of broadband services, including those that do not have a constant bit rate. For example, the Internet connects to many bursty traffic sources, which are best accommodated by ATM SBR Class 2 or GFR, which have less rigid requirements on cell transfer delay and cell delay variation. Mapping these non-real-time services into a fixed bandwidth channel prevents the ONUs on a PON from dynamically sharing the upstream PON bandwidth. For these non-real-time traffic types, the ability to assign bandwidth dynamically is expected to provide higher efficiency than the current static granting mechanisms.
The DBA protocol consists of three strategies, referred to as non status reporting, status reporting and hybrid types. The non status reporting type strategy is invoked by monitoring traffic in the OLT where lengthening buffer queues can be viewed as a requirement for extra bandwidth. The status reporting type strategy is invoked by status reports sent from ONUs to the OLT, where requests for more bandwidth are explicitly given. The hybrid strategy is invoked by both monitoring traffic in the OLT and processing in the OLT the status reports from ONUs. The recommendation does not specify detailed mechanisms or algorithms for these strategies, but rather message requirements and the required interfaces at specified reference points.
|
Written on 3:20 PM by ooe
The most obvious change that can be made to a PON system is to increase its overall bandwidth. This would allow more bandwidth per customer on average which should mean that the cost per unit of bandwidth will decrease.
The early APON standards specified 155 Mbit/s to a maximum of 32 users. This results in a bandwidth per customer in the order of 5 Mbit/s. Whilst this can be considered a great improvement on current systems, and is symmetrical in nature, in practice this bandwidth is not a great deal and would not be capable of adequately supporting the ‘Triple Play’ of voice, data and video services. Furthermore, VDSL can be configured to give bandwidths in this range and therefore the base specification APON is probably not of tremendous interest.
More recently the APON standard has been extended to allow for 622 Mbit/s downstream and 155 Mbit/s upstream asymmetric bit rates as well as a 622 Mbit/s symmetrical option. These are seen as being of greater value, since with the same 32 way split specification these can deliver in the order of 20 Mbit/s to the user. This is closer to the sorts of bandwidth required for a wider mix of services.
Currently FSAN is considering a 1.25 Gbit/s and even a 2.5 Gbit/s PON option, using new architectures not based on their baseline which was G.983.1. These new PONs may have a use in the telecommunications arena probably only as a feeder/backhaul network for VDSL or other access network systems. The IEEE is also becoming active in this area and will soon specify a 1.25 Gbit/s PON, although details are sketchy at the time of writing this chapter.
Point-to-point Ethernet systems are also becoming more prevalent although the lack of standardisation is hindering the creation of consistent product offerings amongst different vendors. Systems comprising of a head-end and street electronic concentrators are being installed, i.e. in Sweden, where they are providing high speed Internet connections to multiple dwelling unit.
|
Written on 3:20 PM by ooe
Recent trends in PON specifications have been of a rather technical nature, looking at the many facets of PON architectures, and attempting to make the systems more efficient or easier to use in the real world where deployment issues can entrap the unwary. This section highlights recent trends in PON system specification but the reader should note that none of these new specifications are mandatory. Indeed they may or may not be used individually on a particular system implementation. This can lead to a myriad of systems, all incompatible with each other, but conforming to the ITU-T recommendations and originating from FSAN.
|
Written on 3:19 PM by ooe
The IEEE hope to capitalise on the success of their existing Ethernet products. There is no doubt that these existing products were successful as they are in widespread use and form the basis for most computer LAN installations. Because of their high rate of adoption, these systems and the components from which these systems are built, are extremely cheap and well suited to the task for which they were designed.
Initially, design effort is being expended on creating a 16 way split PON with a reach of 10 km; however, how this will be used in a real access network is unclear as FSAN believe 32 way splitting, with a 20 km reach is optimal for such networks. The IEEE PON is designed to run at 1.25 Gbit/s, double that of the highest speed existing APON.
The idea is that by using these high volume components in different ways, a new Ethernet PON system could be designed. This is not quite the reality of the situation, however, as the functionality of Ethernet systems is not the same as that required by PON systems, Ethernet based or otherwise and significant changes will be required. For example, PON systems usually require a ranging protocol which enables remote multiplexors to access a common fibre upstream bearer channel with correct timing so as not to corrupt each others signals.
Furthermore, the optical power constraints are more severe as the split ratio and the bit rates of the PONs go up which mandates the use of more expensive optical components.
|
Written on 3:19 PM by ooe
The FSAN group has been around for 17 years now and is split into optical systems and copper systems. This chapter will deal only with the optical systems. The basic FSAN recommendations were input into the ITU-T some time ago with the baseline documents being G.983.1 and G.983.2. Recent work has resulted in new standards which enhance the usefulness of the basic PON systems in certain circumstances. These will be discussed further in this chapter.
What is of interest is that work in the group is not coming to an end as one might expect. Rather the group are moving on to higher specification PON systems. Work is progressing on higher bit rate PONs although the technology underlying the new architectures (i.e. ATM or Ethernet) has yet to be agreed. Limited further information is included later within this chapter as the key agreements have yet to be reached.
|
Written on 3:18 PM by ooe
Optical fibre access network technology has become more complex over the last few years with bit rates steadily increasing. PON systems are available with bit rates in the range 155 Mbit/s to, currently, 622 Mbit/s. Newer systems, both PON and point-to-point, will allow this trend to continue with 2.5 Gbit/s systems becoming practicable in the near future. With this increase in technology has come an increase in the opportunities for fibre system to cost-reduce more of the network, including the metro space – a fact that increases the appeal of the technology. Some of the newer PON systems are being designed using Ethernet as the underlying technology with a promise of reducing the price of the technology still further. How this will occur in practice, however, is still unclear. What is certain is that optical access technologies are expanding and are sure to find an opening in real networks in the foreseeable future.
Developing equipment for the access network has always been a risk for the equipment manufacturers. The size of the network means that equipment volumes can be huge but the equipment is extremely cost sensitive. Furthermore, the sheer cost of equipment rollout means that network operators do not attempt such a major upgrade of their networks lightly. Indeed, to date there have been no ubiquitous upgrades of the UK access network, rather there have been many small, ad hoc changes. Things are now beginning to change with the introduction of ADSL, although this still utilises the most costly and fault-prone parts of the copper access network.
Whilst some new services are being rolled out to businesses that can afford them, these roll-outs are focused on those customers wanting to use the bandwidths immediately with little effort on overall network upgrades. In the UK, services such as the short haul data service can provide upto 2 Gbit/s to customers separated by no more than 40 km, whilst conventional data networks can supply in the order of an STM1's worth (155 Mbit/s) to high bandwidth customers, again mainly business related.
In terms of residential customers, progress in upgrading the network has been slow, mainly because of the perceived lack of value of bandwidth upgrades. The residential market has been getting ever more cost sensitive, with the restrictions in permitted price increases and a glut of bandwidth in the core now hurting many CLEC companies quite significantly. Against this backdrop comes ADSL, enabling users to get 500 kbit/s downstream and 256 kbit/s upstream at an extremely competitive price.
Back in 1989 people predicted the introduction of Passive Optical Network (PON) technology into the network [1]; however, this roll-out did not occur for several reasons, the main one being that there was no perceived market for the types of services that optical fibre technology could bring. This is now beginning to change.
Current bandwidth upgrades will not satisfy customers for ever. There is a question as to when residential customers will require more bandwidth than that offered by ADSL but it seems to be required sooner rather than later. During the last 10 years, there has been a step change in the computing power available to the average residential consumer. Processing power has increased from 8 Mbit/s systems to the current 2 Gbit/s systems (with 3 Gbit/s around the corner) and there is little sign of the pace of technological advancement slowing. This is already leading to customers being able to manipulate video images in real time. What they cannot do is transport this data, in real time, across the telecommunications network.
Following a similar line, point-to-point Ethernet technologies are also receiving attention with a variety of systems becoming available from different manufacturers.
If developments continue at the same pace, then there will be a need for affordable interconnections at speeds currently unattainable by ADSL. Whilst VDSL could cope in the medium term, questions are already being asked about its suitability, given the high operational costs of the required street located electronics and questions relating to radio interference from overhead copper cables.
Against this backdrop the future development of fibre systems for the access network seems to be important. Fibre access network products are already becoming available but most of these are expensive, difficult to implement and are not full service compatible.
Whilst fibre technology has yet to mature enough to make cheap access systems, the major problem is the lack of volume in the sector to date. It is against this backdrop that there is a need to develop affordable, flexible systems. A variety of companies and people, however, are rising to the challenge. It has become noticeable that there are more start-up companies designing and building equipment for this market sector than ever before and that existing major players are now moving into this field with mainstream products.
Whilst standardisation of systems has been shown to lead to a steep reduction in their price, along with increased availability, standardisation too early can lead to poor products. Hence pre-standards discussions in an open environment can be a useful first step. This can be brought about at specialist fora at which the associated problems and challenges can be discussed and indeed such fora have been created and are making good progress in setting initial requirements and solving technical problems.
The best known standards forum is FSAN. Until recently the FSAN group has been the mainstay of optical fibre access system development. Seven operators, including BT, first met in the spring of 1995 in order to begin the FSAN initiative. FSAN stands for Full Service access network and had an initial focus on passive optical network system (PONs). BT had previously been experimenting with ATM-based PON systems [2] but these systems remained largely in-house.
The network operators invited equipment providers to join the group in order to progress activities in a manner suited to the production and specification of such equipment. Using this mechanism the number of contributors to the FSAN forum could be controlled, preventing the organisation from becoming unwieldy. It is still the situation today that equipment vendors only join FSAN by invitation, although they are free to solicit such invitations.
The IEEE are also now active in this area and have their own forum, the Ethernet in the First Mile (EFM) group, trying to create access network standards around existing Ethernet standards.
These groups will now be discussed in some detail.
|
Written on 2:32 PM by ooe
Overbuilding refers to the situation where a new network is being provided in addition to established infrastructures such as twisted pair for telephony and coaxial cable for cable TV services. The incumbent networks may be old and the ability or will to deliver broadband data services is weak. The opportunity for the over-builder is to offer broadband data as an alternative or in addition to the other services. New-entrant operators are introducing point-to-point Ethernet systems to satisfy the demand for data in both residential and small to medium sized business sectors.
The standard PON is based upon ITU G983 series and is capable of delivering 622 Mbit/s symmetrical services by circuit emulation, or with ATM or Ethernet interfaces. These PONs are currently focused on the business market but work is in progress to widen the service set to include telephony and cable TV services for residential green field applications. The cost of fibre access is high, so it is a challenge for any fibre access solution to be profitable in the residential overbuilding scenario. A data-only solution is likely to be simpler and cost less than a full service access network but has fewer revenue streams available to pay for its introduction.
A data-centric approach is possible using point-to-point fibre and LAN technology to connect buildings and apartment blocks to service providers. An active node such as a hub, switch or router at the fibre entry point to the building is used to connect individual offices or residences via category 5 cable (one twisted pair for each direction of transmission). This type of network is gaining popularity but will struggle to deliver high quality telephony and real time video for residential applications because of the absence of circuits. Contention for bandwidth in such networks is difficult to control if the access protocol remains IP.
The G983 series PONs are able to deliver data-only services and multiple line digital telephony by circuit emulation. These systems can be considered as an alternative to SDH and PDH point-to-point rings for overbuilding in a business community where demand for telephony in blocks of up to 24 (T1) or 30 (E1) channels are commonly found.
|
Written on 2:32 PM by ooe
As mentioned in, green field (new-build) is a key problem area, which gives a focus to the problem of FTTH. New homes will incur the costs of civil works (digging) and outside plant anyway so the issue is which technology to install: twisted pair, coaxial cable or fibre, or a mixture of all three.
Green field fibre construction has been neglected because of the advance of broadband over DSL, cable modems and UMTS. Nevertheless, on a worldwide basis, there are significant volumes of new homes being built and even more homes requiring basic telecommunications services in developing nations.
The problem for system designers is to provide multiple-service capability: telephony, video and data. A rich service set allows a single fibre network to compete more effectively in economic terms with twisted pair and coaxial cable as the traditional means of delivering each of these services. None of these competes with fibre in terms of capacity and fibre offers more resistance to water ingress, lightning damage and electrical or RF interference. The opportunity for fibre systems vendors and operators is to make a single fibre more profitable than a bundle of wire pairs and coaxial cable for the delivery of multiple services. The full services access networks (FSAN) group was set up with this as the final goal. The activities of this and other standards bodies will be described later.
Providing traditional telephony services is troublesome over fibre because of the absence of wire for power feeding. Fibre is dependent upon the local electricity supply and batteries for back-up. Nevertheless, since this problem has been solved for cellular telephony, similar solutions can be used for fibre without incurring the additional cost of a copper access network.
There are numerous ways of providing video services. Both broadcast and switched are possible via PONs. The fibre infrastructure may carry broadcast services in all legacy modulation formats but time division multiplexing (TDM) offers opto-electonic simplicity and high power budgets. Pure TDM unfortunately is not a format used for video services in the mass market yet, although digitally encoded channels in TDM ‘bouquets’ are found in FDM broadcast systems. Such systems may come a close second best for transmission of broadcast video over fibre. Base-band switched video services direct to a PC are emerging via the Internet.
In the UK, although the franchises for cable operators expired in 2001, most green field build has remained a combination of twisted pair and cable. Twisted pair is provided to almost all locations, while cable is typically provided in the less rural areas. Fibre to the home has remained a niche application because of the continuing success of ADSL and the higher cost of fibre as explained earlier. Only in a few select trial areas has fibre to the home been deployed.
For BT, green field build has been a relatively low volume offering a limited opportunity for new infrastructure. However, there is expected to be a need for around 1 M new homes in the UK over the next decade with large developments in SE England.
A green field solution used by BT is TPON (telecommunications over a passive optical network) in remote cabinets or buildings. It has been narrowband only because of restrictions on cable TV services and an uncertain market for interactive broadband. In most cases the feeder to customers is copper pair rather than fibre.
So, despite the competitive environment that has been opened up since 2001, fibre to the home systems have remained difficult to justify, even in green field where the infrastructure costs are always encountered. This may change in the future as homes require data rates in excess of ADSL capabilities and more broadband services become available.
In the USA, the situation is similar; Telcos are beginning to compete with cable for multiple service provision. This competition is making incumbent operators think about radical restructuring of the access network to offer FTTH particularly in newbuild situations and there are a number of trial areas where FTTH has been used.
In mainland Europe and Japan cable operators and Telcos have followed separate evolution paths with separate licenses for cable TV and telecommunications and little competition between the two.
|
Written on 2:31 PM by ooe
The choice of deployment strategy will determine how much and at what stages capital is spent, the coverage and how much infrastructure is provided. The choice of strategy therefore represents a key economic consideration for fibre access. Three approaches are considered here: green field, overbuilding and upgrading an existing network.
|
Written on 2:31 PM by ooe
Alternative routing is required by some customers to allow service to continue in the event of a single or multiple failures. Resilience is also required when the traffic concentration is high.
Topologies that offer route protection include point-point or PON systems with two separate fibre paths, dual rings with loop-back protection and mesh networks. All of these networks are used but are not being considered for mass-market fibre to the home systems (FTTH) because of cost.
|
Written on 2:31 PM by ooe
Rings are used extensively in core and metropolitan networks because they offer a means of interconnecting a number of nodes (cities, towns) in a way that enables large capacities to be transported with resilience. The opto-electronic technology is the same as that needed for point-to-point unless WDM is involved. If a route fails another route can be found by sending traffic in the opposite direction round the ring as shown in.
The amount of fibre needed is considerably less than point-to-point networks, however there may be more civil works involved in serving a community as two fibre entry points are needed if strict route diversity is required. Furthermore if the ring is to be used to provide resilience, two optical paths and two sets of transceivers are needed to offer loop-back protection which increases the cost. Without loop back protection a unidirectional ring is vulnerable, as a single failure causes all of the traffic to be lost.
|
Written on 2:31 PM by ooe
Passive star architectures can be implemented with either passive power dividers or with wavelength division multiplexing (WDM) components. Both offer a one-to-many optical concentration function. The passive optical network (PON) approach, shown in, makes use of passive power dividers in the network and promises lower cost and more flexibility for early deployment. It may use any available multiplexing schemes in the equipment at either end of the network. WDM on the other hand promises a lower power budget and is therefore capable of giving more bandwidth per user on a dedicated basis, rather like a point-to-point fibre. The PON or tree architecture allows many signals to be combined onto a common bearer into the central node by suitable choice of multiplexing rather like conventional cable TV over coaxial cable. Traffic may be concentrated at any point in the network. In general, costs are minimised if the concentration point is near the customers. In the downstream direction, the multiplex is broadcast to all users. In the upstream direction broadcasting does not occur but a multiplex may be formed optically ‘on-the-fly’ using multiplexing such as WDM or TDM in the terminal equipment. The multiplex must be designed to avoid mutual interference, which would arise if channels are chosen at random.
|
Written on 2:30 PM by ooe
Point-to-point systems use one or two fibres to connect a pair of nodes. The full bandwidth of the fibre is available for future upgrades and a high power budget is available for long-distance transmission. A drawback is that the network is not well-suited to broadcast services.
As the number of nodes in a network grows it is desirable to concentrate traffic to avoid the huge proliferation of fibre, which would arise with an unlimited mesh size. The choice of concentration point will affect the cost of the network. For example, if the concentration is nearer the customer the amount of fibre required can be reduced. However, point-to-point systems require a concentration point with active electronics. Consideration of environmental issues such as powering and temperature control and water ingress is then needed. These issues can be solved more easily if there is an existing building such as a central office offering environmental control. In the green field or overbuild situation these buildings may not exist and an external plant solution such as a pedestal or cabinet will be required at some cost.
The benign environment of an existing central office can perform this concentration function for point-to-point systems. The location of this building has been chosen to suit the needs of twisted-pair transmission. With fibre access using point-point technology, it is almost inconceivable to replace the wire pairs with fibre on a one-for-one basis because of disruption to existing services. Alternatively the installation of a new cables as an overlay containing massive numbers of fibres is very costly
For the mass market, operators are considering alternative architectures than traditional point-to-point for the local loop – the passive star that avoids the cost of active concentrators and can reduce opto-electronics costs by resource sharing.
|
Written on 2:30 PM by ooe
The choice of topology is critical because it affects the cost of the network, the services, and the reliability of the access network. Once installed it will be very difficult to change the physical configuration, so topology needs careful consideration. Here we examine the key candidate topologies: point-to-point, passive star, and ring.
|
Written on 2:30 PM by ooe
Bearer service requirements include telephony, data (both guaranteed and best-effort) and broadcast TV. For business users high availability is a key requirement whereas a broadcast TV service may be of secondary importance. For the residential community a rich set of switched and broadcast video services will be attractive but lower availability will be acceptable if it saves cost.
|
Written on 2:29 PM by ooe
Demand for bandwidth is the key driver for fibre in the access network. Fibre has served the large bandwidth needs of long-haul networks for 20 years. Businesses or buildings, which can concentrate many users' traffic onto a fibre, have also been directly connected. The fibre access systems used have much in common with core transmission techniques such as PDH or SDH and are configured as point-to-point or ring networks. Corporate access customers have traffic concentrations, security and reliability requirements similar to core networks.
The big challenge is to solve the problem of the mass-market with the ultimate goal of fibre to the home (FTTH). Much of the technology needed has been around since the mid-1980s, but until now FTTH has been achieved only in technology trials or small-scale deployments. For FTTH to become ubiquitous, costs need to fall to be comparable with alternatives such as twisted pair or hybrid fibre-coax. Three important costs are the technology, installation and ownership. For FTTH to succeed these costs must be lower than the expected income from services over the life of the system.
Green field (new-build) is a key problem area, which gives a focus to the problem of FTTH. The cost of providing any fixed access network is very high when the civil works are included. New homes will incur these costs anyway so the issue is whether to deploy twisted pair, coaxial cable or fibre, or a mixture of all three.
Fibre promises capacity far in excess of the alternatives and offers future-proofing as demand for broadband services grows. What is needed is an entry-level system, which is cost-effective to justify the investment in a fibre infrastructure, rather than an alternative. To do this it must be comparable in cost and carry a richer set of services than the alternatives. Once installed the return on investment of the fibre network over its whole life could be higher than the alternatives, which would not compete so well as the demand for bandwidth increases.
This chapter sets the scene for a mass-market product by focusing on fibre access technology choices that influence cost and/or service capability.
|
Written on 4:17 PM by ooe
The development of automated interfaces between operators is only part of the OSS challenge. Many changes have to be made to the backend (back office) systems of both operators. It is well understood in the IT industry that changing legacy systems is complex, expensive and time-consuming.
The LLU systems development has had to support many new products types. Examples from the UK design include:
-
exchange area data that relates exchange servicing areas to postcodes and makes the information available securely on a web server;
-
collocation facilities – this requires the linking of BT forecasting and planning processes to LLU operators' plans so as to provide and reserve frame, tie cable and collocation capacity;
-
appointment reservations systems for LLU service provisioning and repair including access to exchange buildings;
-
provisioning rules for job management systems.
In some cases the customer may wish to transfer (or port) their telephone number. This required some integration of the processes and systems for number portability products with those for LLU products. The allocation of a line to a LLU operator requires the suppression of the automated billing which would otherwise be sent to the retail customer for an ‘in service’ access network path.
Considerable changes have been needed to the repair processes to cover demarcation of responsibilities between operators, co-ordination of testing, and escalation processes with the consequential impacts on the BT maintenance and repair systems. Substantial modification to the process and systems for testing of lines has been necessary.
New management information systems are needed to track statistics for LLU provisioning and repair. These are needed to monitor and track actual service performance and support the provision of information to the regulator. These requirements have resulted in significant new developments and hundreds of changes to existing legacy systems in an unprecedented short interval of time.
Several examples have been given derived mainly from the authors' experience in BT. These services, together with wholesale DSL products, allow competitive operators to use existing metallic access networks belonging to other operators without investing in building their own networks. Implementing LLU involves complex developments of technology, support systems and processes. The work requires agreement between network owners, regulators and representatives from other operators. BT was among the first operators to fully meet the challenge of the EU LLU regulation. The first collocation areas in the BT network were handed over to operators in late 2000, and in January 2001 BT released its first unbundled loop. The next few years will see a massive expansion in the availability of broadband access services and there may be volume demand for LLU products when the telecom sector recovers from the downturn which started during 2000. However, a wide range of broadband access technologies is available including DSL, broadband radio, satellite and cable TV. Only time will tell what role local loop unbundling will play in this expansion. The eventual extent and rate of unbundling depends to a very large extent on the commercial interests and firm orders received from new entrant operators.
|
Written on 4:16 PM by ooe
The definition of industry processes for LLU was an essential precursor to the definition of the technical interfaces. The LLU Automation Forum (LLUAF), a working group established under the OPF, was set up to address this specific need. An innovative approach was adopted for the specification of these interfaces.
Rapid progress had been made in the development of business-to-business e-commerce technologies and supply chain integration. It was recognised that OSS integration might be simplified if they were employed. The ordering interfaces were based on the Commerce One Common Business Library CBL. This is a recognised e-commerce library that allows the specification of purchase orders and order status using the XML (extensible mark-up language). The library comprises a large number of definitions of general use in constructing ordering processes and messages. For example, CBL defines a structure of a purchase order and semantics for data types such as customer name, address, order reference codes and the semantics of delivery and order dates [8]. The Commerce One library needed some adaptation to make it suitable for use in telecommunications services. These extensions are based upon conclusions reached in a research study in an EURESCOM project P908 on business-to-business OSS gateways [9].
Publication of the LLU message sets for the agreed industry processes was completed in January 2001 and industry feedback and comment was solicited through the LLU Automation Forum and the UK Telco API Forum web sites [10]. Promotion of the work through the ITU-T, and the Tele Management Forum lead to world-wide acclaim of the leading-edge innovative approach being taken by BT and the UK industry.
|
Written on 4:16 PM by ooe
The quality of service experienced by the end-user is highly dependent on the ability of all parties to provide, maintain and bill for LLU services in a cost-effective manner at high volume levels. Close co-operation between operators is required both at initial service provision and during any maintenance activity that may be required. It was recognised that a high degree of automation would be required to support large volumes of loops. This required that LLU operators be able to connect their operational support systems to BT's computer systems which were not designed to support LLU. Over 20 major computer systems were impacted and the integration task was one of the most complex undertaken since BT first installed customer care and billing systems in the 1980s.
|
Written on 4:16 PM by ooe
Processes were developed for line sharing and sub-loop LLU. Line sharing adds complexity to many LLU processes as two service providers use a single access line. New scenarios are introduced making the whole operational environment significantly more complex. Operators must now deal with requests for new shared lines, conversion of a telephony line to line sharing and reversion back to their original state, or a complete cease of all services. When a fault occurs there is potential for confusion. For example, it may be unclear whether the fault is on the network owner or LLU operator part of the line; the end-user may not know which operator to report the fault to and each operator may consider that the fault lies on the other's network. Processes and procedures are required to resolve any issue quickly.
|
Written on 4:16 PM by ooe
Accurate and timely handling of customer's reported faults is fundamental to achieving customer satisfaction. LLU services rely on a combination of network elements including the metallic path provided by the network owner, additional components supplied by the LLU operator and perhaps some components supplied by the end-user. When the end-user experiences degraded service, it may be unclear which component is the cause.
In the UK, the OPF group developed processes to reduce the scope for multiple handing of the fault report and to support efficient and timely resolution of the fault report. In full unbundling the LLU operator is responsible for receiving fault reports from the end-user and should pass them to BT only if the LLU operator has reason to believe that a fault exists on the BT network. The situation is more complex where line sharing is used as both BT and the LLU operators are providing services to the enduser using the same metallic path. The end-user should report service degradation to the appropriate service provider in the first instance. Maintenance and repair activity on one service is likely to disrupt the other service so close co-operation between BT and LLU operator may be required. BT already has systems and processes to deal with problems affecting the narrowband (telephony) service due to network faults. These processes have been adapted to enable repair engineers to return the unbundled loop to specification based upon electrical performance. Once a repair is complete the unbundled loop is returned to the LLU operator with a report of the relevant electrical characteristics as proof of repair. Furthermore the LLU operator can purchase further test products to enable it to obtain additional information about a metallic pair. The LLU operator is, of course, solely responsible for dealing with faults caused by that operator's equipment or in the operation/configuration of the broadband service by the end-user.
|
Written on 4:15 PM by ooe
Provision of metallic pairs is carried out in three stages: the pre-order enquiry, deliver customer order and order acceptance. At the pre-order enquiry stage the operator is provided with information about the customer access line. The operator may wish to have a new metallic pair or to take over an existing one. The line information enables the LLU operator to decide if its service will work over the metallic pair to be provided. The operator may decide not to go ahead with the order if its service will not work because of the distance of the end-user from the exchange. If the operator decides to proceed, BT provides the pair as requested and the operator has two days to decide to accept or reject the pair. This period provides sufficient time to enable the operator to test its service.
The provision of the pair is usually quite straight forward. However, a number of issues need to be considered. The end-user is now a customer of the LLU operator and BT has to rely on the operator to make arrangements for access to their premises. Further complications are introduced if an order is combined with other products such as number portability, carrier pre-selection and/or calls and access.
|
Written on 4:15 PM by ooe
Once registration has been completed the LLU operator can order physical collocations or distant locations from the BT exchanges in the geographical areas in which they are interested. At each intended point of presence, a two-stage survey and building process with identified timescales was agreed. This was necessary to establish an audit trail should queries arise.
When LLU was initially launched more than 40 operators registered an interest and there was clearly potential for a large volume of collocation orders. The OPF group agreed that a process was required to manage this bow wave of orders before demand settled down and business as usual processes could be adopted. A demand management process was devised, which was to be operated by the Electoral Reform Society and Oftel, to decide in which order BT would tackle equipping its exchanges. This process was put in place in September 2000. In fact the volume of orders was massively below the industry forecast level and BT was able to handle them without using the bow wave. From December 2000, LLU operators were able to order distant location on a business-as-usual basis. Based on a fuller understanding of what was practicable, in February 2001, BT announced that it could handle demand for collocation space as business-as-usual.
|
Written on 4:15 PM by ooe
The gain business process covers registration for LLU service with BT. New contracts have been developed which cover the provision of network information required to enable operators to develop their business plans, and the ordering and maintenance of new LLU products from BT. If operators decide to go ahead with their business cases then registration is completed and BT will set up service accounts for billing purposes.
|
Written on 4:15 PM by ooe
Like any large network operator, BT has a large number of complex processes and systems which govern the operation of its access network.
However, it is a starting point from which to determine what would need to be modified. Typical examples of these changes include:
-
how appointments are made with the end-user for equipment installation by BT;
-
how faults are determined and resolved (that is, whether a fault is in the LLU operator's service area or on the network provided by BT);
-
what to do if the end-user contacts BT directly rather than contacting the LLU operator.
These and similar issues were discussed at several OPF meetings which resulted in the interface between BT and the LLU operator being defined. Although this work was completed in April 2000, industry and Oftel requested numerous changes resulting in major reviews. A simplified end-to-end description of the unbundling process is shown below.
-
Operator registers for LLU service with BT.
-
LLU operator orders points of presence (collocation or distant location).
-
BT or contractor builds points of presence.
-
LLU operator installs equipment.
-
LLU operator requests unbundled loop.
-
BT provides unbundled loop.
-
LLU operator provides service.
-
LLU operator reports faults to BT.
-
BT investigates problems.
A number of activities such as billing and field force control are inherent in these steps – threads running through every major process. Add to this the number of scenarios identified for unbundling (over 100) and the scale and complexity of the operational environment starts to be seen.
|
Written on 4:14 PM by ooe
LLU operators in many countries have lobbied to be allowed to place equipment in the same rooms as the network owner. They argue that this is the best way to ensure parity of treatment between operators. This raises significant security concerns as local exchange buildings were almost always designed and built for sole use of a single network operator. Cageless collocation is no simpler to implement than physical collocation as provision must be made to connect power services, metallic pairs and backhaul services to the LLU operators equipment. Design calculations to ensure that sufficient air conditioning capacity are more complex in a cageless environment. Safety and security issues are more significant where the LLU operator shares a room with the network owner. In many cases the LLU operators' equipment and people have to satisfy more stringent testing or vetting before being allowed into the equipment rooms.
|
Written on 4:14 PM by ooe
Distant location was initially conceived to address the situation where some BT exchanges would offer insufficient accommodation to meet OLO demands for physical collocations. It uses an external tie cable from the BT MDF to premises or street furniture obtained by an LLU operator, near to the BT building. Providing accommodation for distant location is the responsibility of the LLU operator although BT may be sub-contracted to undertake the installation.
Distant location does not require the LLU operator to place equipment in or even to enter the BT building so there are many fewer restrictions on the design of the LLU operator equipment and operations. Distant locations have been built in nearby buildings and in self-contained street cabinets.
|
Written on 4:14 PM by ooe
Physical collocation allows LLU operators to place equipment in local exchange buildings and allows their authorised people to gain access to relevant parts of the buildings to operate their networks.
Providing collocation facilities means much more than simply making floor space available. Operators will require power, ventilation and related building services. For security reasons some degree of segregation is required between LLU operators' equipment and network owners' equipment. The agreed design criteria for physical collocation includes:
-
LLU operators must be able to source their equipment from the global marketplace.
-
A wide range of rack and cabinet sizes to be offered as well as differing power and environmental operating characteristics.
-
Different and innovative deployment models were required as each LLU operator would be competing for business within each exchange area. These models might range from a small ‘toe in the water’ presence to an installation able to serve several thousand customer lines from the outset.
-
No ‘one size fits all’ constraint.
-
Recognition that some LLU operators would wish to have a dedicated room tailored to their stated need, offering a reasonable level of physical security.
-
Some operators might wish to collaborate and share a collocation space in an area.
An LLU Hostel product was designed which formed the primary means of responding to demand and meeting the outlined criteria (Figure 7.6). In essence, the LLU Hostel enables an LLU operator to simply order one or more equipment bays – an area of floorspace suitable for equipment racks. Electricity is provided to each bay via an AC final distribution fuseboard providing a designated maximum power load. Expectations that LLU operator demand would be focused on exchange buildings serving the main population centres led to an initial minimum hostel size of two suites. This lower limit was subsequently dropped and a reduced version of the product was introduced to satisfy demand for hostel bays in smaller exchange buildings or where space constraints lead to the adaptation of rooms not originally designed for housing network equipment. Key design features of the LLU Hostel range of collocation spaces includes:
-
The use of forced air cooling will generally be made.
-
An HDF located at, or close to, the end of each bay. The HDF is used as the interface for internal tie cables provided by BT to the MDF.
-
An initial provision of tie pairs to the exchange side of the MDF. Additional ties can be ordered separately.
-
Fire/smoke detection.
-
Access control. BT has modified its building security systems to allow the other operator's people to access the relevant parts of the buildings without needing to be escorted by BT people.
-
An AC power distribution board from which LLU operators take their power feeds.
-
A cable support system using ‘unistrut’ ironwork in a simple configuration. The end-product is a ‘ready to go’ environment into which an LLU operator can readily deploy and commission equipment.
|
Written on 4:13 PM by ooe
Collocation is considered to be one of the most important services in LLU. LLU operators must place equipment close to the MDF to which they wish to connect. There are three broad types of collocation which are physical collocation where the operator rents space inside the local exchange building, distant location where the operator's equipment is housed in a nearby building or street cabinet and cageless collocation where equipment belonging to the LLU operator is located in the same room as that of the network owner.
In the UK, BT has more than five thousand local exchange buildings and these were never designed or intended for third-party access. When designing collocation products BT studied the similar products offered by other incumbent operators in the USA and continental Europe and adopted the best practices from these operators.
The introduction of collocation requires that standards for equipment and operations be agreed. In the UK the standards for equipment and the inter operator processes for collocation were developed and trialled by BT and representatives of the LLU operators in a sub group of the Operator Policy Forum during 2000.
A key aim of the product design was to minimise the need for equipment specifically for designed or modified for LLU and wherever possible off the shelf hardware was used to minimise supply chain bottlenecks. Statutory requirements and customer affecting considerations such as safety, fire detection, planning law, security and risk management all had to be satisfied. In addition, BT's operating licence places it under an obligation to secure its network equipment.
The first collocation rooms and distant location facilities were completed in December 2000. At the completion of the collocation trial BT was well placed to design and build substantial volumes of collocation facilities in response to firm orders placed by LLU operators.
|
Written on 4:13 PM by ooe
The technical specification of the unbundled line forms a part of the contract between network owner and the LLU operator. It is important that the network owner is able to test the lines before provision and during maintenance activities. BT has invested heavily in test equipment for use on its conventional telephone lines and it was logical that the same equipment should be used to test unbundled lines. The connection between the test equipment and the metallic line is a part of the local exchange concentrator but this is not connected to unbundled lines. The short term solution to this problem was to provide a physical test point connected to the unbundled metallic path near the point where the tie cables were connected to the MDF. When a test was needed on an unbundled line a BT engineer had to visit the MDF and make a manual connection between the line test systems and the metallic path test point. A further visit was required at the end of the test period. This solution was simple to implement but relatively slow and manually intensive to operate.
BT has decided that the longer term solution to the test problem is to connect all the unbundled lines through a specially provided test access switch matrix. This device simply automates the process of connecting the test equipment to the line. This could not be adopted at the initial launch of LLU products owing to the time taken to specify and procure a switch matrix and in particular to integrate it into BT support systems and processes. The test matrix will allow tests to be carried out more quickly than the manual test solution allows.
This test solution is appropriate for both full unbundling and line sharing. It is not appropriate for sub loop unbundled lines which do not connect to the local exchange site. A different approach must be developed if these lines are to be tested without manual intervention.
|
Written on 4:13 PM by ooe
Cost effective and timely maintenance is an important component of any telecommunications product. This can be difficult to achieve in the LLU situation as responsibility for maintenance is divided between two or more operators. Clear demarcation points between areas of responsibility are crucial to effective transparent processes and to ensure a good customer experience. In the UK the NICC DSL Task Group addressed this concern early in the development of LLU. The unbundled loop is defined between two demarcation points located at the network termination point in the customer premises (usually a telephone socket) and at the handover distribution frame in the LLU operator room or distant location. A clear and unequivocal specification of the performance of the unbundled pair was written and agreed by the DSL task group. It was expressed in terms that could be measured by either BT or the LLU operator and took account of the skills and tools available to the workforce. The specification provided a base from which DSL Task Group developed a code of practice for maintenance activities that allowed each operator to understand their responsibilities under fault conditions.
|
Written on 4:13 PM by ooe
LLU will often be used in conjunction with DSL technology to provide data services. BT has, for many years, been a leader in research into DSL crosstalk interference and in the formulation of deployment rules to control it [4,5]. Before LLU was introduced to the UK, BT was solely responsible for ensuring that its transmission systems were used in a way that avoided excessive crosstalk on its network. With the introduction of LLU, it was no longer appropriate nor possible for BT alone to control system deployments. A new set of rules was required that all operators could follow. The new rules had to be clear, easy to implement, fair and had to take account of likely future developments in the access network.
The NICC DSL Task Group [7] was the cross-industry technical forum that was tasked by Oftel to design an access network frequency plan to meet these requirements. The task group was chaired by BT and was open to all interested parties. A wide range of telecommunications operators and equipment vendors actively participated with BT and Oftel. The access network frequency plan (ANFP) for the BT network is described in detail in Chapter 4.
The ANFP for the BT network controls crosstalk by defining limits on the power, frequency and location of DSL systems. The ANFP works by placing limits on the worst-case crosstalk noise in the network. By ensuring that new developments in the access network are no more harmful to their neighbours than current systems, the ANFP provides the stability that allows all operators to plan for the future with confidence. The ANFP places the onus on individual operators to ensure that their DSL deployments are compliant and remain so. The DSL task group designed a code of practice to be used when investigating allegations of interference.
When the ANFP was first mooted, line sharing and sub-loop LLU were not required so the first issue of the ANFP made no specific provision for them. Similarly, international standards for VDSL technology were at that time not agreed and the first plan excluded VDSL. At the time of writing the DSL task group is extending the plan to accommodate VDSL and sub loop unbundling.
|
Written on 4:12 PM by ooe
This section describes four of the most significant challenges that BT addressed in its LLU implementation:
-
the design of a plan to control crosstalk;
-
implementation of physical collocation;
-
modifications to the BT operational processes;
-
the design of computer interfaces between BT and the other operators.
Incumbent operators in other countries have faced similar problems but the detailed solutions may be different.
|
Written on 4:12 PM by ooe
In sub-loop unbundling, the LLU operator connects to the loop at a sub-loop access point located between the MDF and the end-user premises). Sub-loop access is suitable for use in conjunction with a high-speed DSL technology such as VDSL [6]. Sub-loop LLU lines may be shared or dedicated. Under shared sub-loop access the incumbent will provide an analogue telephone service to the end-user while the OLO provides a data service.
Collocation is not relevant to sub-loop unbundling as access points are located away from local exchange buildings. To use sub-loop unbundling an LLU operator must provide a street cabinet close to the sub-loop access point and connect it using a tie cable. This is analogous to distant location for classic LLU.
|
Written on 4:11 PM by ooe
If ADSL technology [4,5] is used it is possible for an unbundled loop to be shared between incumbent and the LLU operator. In shared access an LLU operator provides an ADSL-based data service to an end-user while BT provides analogue telephony using the same access line. Line sharing requires extra tie cables and wiring to be provided.
|
Written on 4:11 PM by ooe
The basic form of local loop unbundling, sometimes called full or classic LLU, involves leasing the whole capacity of the access pair to another operator. The LLU operator can use the access pair to provide a range of services. The services offered are limited only by a few rules designed to protect the integrity of the network. The network owner remains responsible for the maintenance of the unbundled local loop from the network terminating point in the end-user premises to the connection point to the LLU operator equipment at the HDF or at a joint in a tie cable.
|
Written on 4:11 PM by ooe
The two major variants of LLU are full unbundling and sub-loop unbundling. Each of these variants may be offered exclusively or with shared use. Full unbundling involves leasing the use of a complete access pair from a local exchange site to the end-user premises. The LLU operator may choose to make exclusive use of the line or may choose to share it with the incumbent operator. In shared access, the incumbent uses the lower frequency part of the line to provide conventional telephony service, while the LLU operator uses ADSL technology to provide a data service.
In sub-loop unbundling the LLU operator is given access to the line at a sub-loop access point which will usually be located at a flexibility point between an exchange building and the end user premises. This may be attractive to operators who wish to use very-high-speed DSL or VDSL technology. If the sub-loop access is shared, the incumbent will use the lower-frequency part of the loop to provide a conventional telephone or basic rate ISDN service.
LLU operators will usually require certain ancillary services to allow them to make use of unbundled loops. A key service is space for the operator to place its network equipment. In the case of full unbundling, the LLU operator may be able to lease some space within the exchange building; this is called collocation. Alternatively, the operator may be able to place its equipment in a nearby building or street cabinet. This is called distant location.
The LLU operator does not need physical access to the main distribution frame. Tie cables are used to connect a handover distribution frame (HDF) in the LLU operator accommodation to the MDF. If distant location is used, the tie cable may be provided by either incumbent or by the LLU operator in conjunction with the incumbent.
Close co-operation between the LLU operators and network owner is particularly necessary during the ordering and maintenance stages. Much of this interaction is being automated using operational support systems that can be connected to the LLU operator's systems. The approach to systems is described later in the chapter.
LLU operators also need core network transmission links to connect the unbundled loops to the rest of their network. These links may be leased from any suitable operator or may be built by the LLU operator themselves. The LLU product variants are described in more detail below.
|
Written on 4:09 PM by ooe
All developed countries have near ubiquitous metallic access networks and they represent one of many Telcos' largest and most valuable assets. Once limited to providing voice access and voice band modems, the metallic access network is now capable of carrying multi-megabit data services thanks to the introduction of digital subscriber line (DSL) technology made possible by recent developments in micro electronics and in digital signal processing. Each customer access line consists of a pair of wires running from the customer premises to the nearest local exchange. The line consists of a number of elements that provide connection, jointing and cross-connection functionality. The key architectural features of BT's metallic access network in the UK are shown. In the UK, access network cables terminate on a main distribution frame (MDF) within the local exchange building and this represents the cross-connection point between the external network and transmission equipment. There are flexibility points in the external network at primary cross-connection points (PCPs) usually located in cabinets seen in the street and at distribution points (DPs) located close to end-user premises, typically on telephone poles. The wire between the DP and end-user is known as a customer feed or customer drop. Access network cable sizes range from 10 pairs to 4800 pairs. In general, cables near a local exchange building have larger pair counts and smaller conductor sizes. Conversely, nearer to customers, the network cables have smaller pair count and larger conductor diameter.
|
Written on 4:09 PM by ooe
If successful customer service is to be provided using LLU, it is vital that incumbents and new entrant operators work closely together. The best way to achieve this is for the operators to work together when designing the processes and any supporting operational support systems for the new services. At an early stage of the UK implementation of LLU a number of working groups were formed involving BT, Oftel and representatives of the many new entrant operators. Each working group had to negotiate a difficult path to achieve commercial agreements between BT and its competitors, while maintaining inter-operator consensus. Two broad streams of work were initiated:
-
A technical task force was set up under the auspices of the Network Interoperability Consultative Committee (NICC) to develop technical specifications for the new products:
-
a plan to control crosstalk interference in the network;
-
technical specifications for the unbundled loops;
-
a code of practice for maintenance of unbundled loops;
-
a code of practice to deal with alleged cases of crosstalk interference.
-
Definition of the processes and support systems was delegated to the Oftel sponsored Operator Policy Forum (OPF). This group had to produce a number of complex agreements including:
-
exchange area data to support LLU operator planning processes;
-
processes to supply LLU service to specific customers;
-
operations and maintenance processes;
-
testing processes supporting both provisioning and maintenance;
-
billing processes.
An essential part of this work was to document and agree the business policies to be adopted by all parties to support LLU services.
|
Written on 3:10 PM by ooe
The duplexing scheme is the method employed to achieve two-way transmission on a single pair of wires. Like ADSL, VDSL uses frequency division duplexing (FDD) to achieve two-way transmission on a single cable. FDD was chosen over time division duplexing (TDD or ‘ping-pong’). TDD was seriously considered for VDSL because the proportion of capacity allocated to upstream and downstream can be easily varied in software. But to operate correctly, TDD modems sharing the same cable must all be synchronised to avoid severe crosstalk. It transpired that this is a serious disadvantage in the unbundled environment because adopting TDD would have required the industry and regulators to agree a synchronisation plan in each country as well as a frequency plan (which would still be required) – not an attractive prospect.
This was the first-ever DSL standards decision to be influenced by regulatory factors. Now, all key DSL standards decisions must take into account regulatory implications – a new dynamic which came into play again in the choice of VDSL frequency plan.
|
Written on 3:10 PM by ooe
VDSL and ADSL modems use similar transmission techniques but the frequency of VDSL operation, and hence capacity can be much higher because VDSL has been optimised to operate over much shorter cables. The two key factors influencing the design and performance of VDSL are the duplexing scheme, and the frequency plan.
|
Written on 3:10 PM by ooe
High capacity transmission systems such as ADSL and VDSL operate at much higher frequencies than the voice telephony which the cables were designed to support. But when BT embarked on development of broadband capability in the early 1990s it was discovered that little research had been carried out anywhere in the world to quantify telephony cable performance at frequencies above 1 MHz. And yet to commit to huge broadband investment without being able to accurately predict the transmission performance of the modems was judged to be unacceptable. BT therefore initiated the most comprehensive programme of technical investigations ever carried out for DSL transmission systems by any network operator in the world.
These technical investigations included:
-
A comprehensive network survey involving high frequency transmission measurements and recording the physical condition of over 7000 telephone pairs throughout the UK selected to be representative of the network as a whole.
-
High frequency crosstalk measurements on 70 live 100-pair cable infrastructures throughout the UK – a task previously thought impossible.
-
High frequency impulsive noise measurements and analysis.
-
Investigations to assess unwanted emissions from ADSL and VDSL systems
Finally, there has been an extensive and ongoing programme of laboratory and field testing with the assistance of modem vendors, to relate knowledge of real network characteristics to modem performance to predict ADSL and now VDSL service capability.
The BT measurement work to verify DSL service capability has to date not been mirrored anywhere else in the world. For example, it was discovered that far-end crosstalk (FEXT) is actually significantly higher in the UK network than had been predicted by extrapolating previous data. This vital information is built into UK planning rules for DSL systems. Moreover, BT's measurement data underpins the UK's Access Network Frequency Plan (ANFP) – the regulatory framework which all UK operators must obey in their DSL deployments to avoid chaos.
The practical effect if the vital information on network transmission performance and crosstalk had not been discovered, would be that high rate DSL systems deployed in the UK could have become progressively unreliable, and even fail, as more and more systems were deployed, with disastrous consequences for operators and customers alike.
In recognition of the significance for the UK of this work, BT was awarded the IFE/National Physical Laboratory Measurement Prize in 1998.
The BT work has been selectively published to influence key international standards. Not surprisingly, the findings have stimulated urgent measurement work in other countries to verify that their broadband deployments are in fact sustainable in the long term.
|
Written on 3:09 PM by ooe
The external factors that need to be taken into account include radio frequency interference (RFI) – typically originating from high power broadcast transmitters; and impulse noise interference – typically originating from transport systems, home appliances and of course the weather. There is nothing that can be done to eliminate these external factors, and VDSL systems must therefore be designed to withstand their influence to operate reliably. Key enablers being robust error protection techniques, and ensuring that the systems are deployed with adequate head-room above the noise – known as the noise margin.
Determining the appropriate noise margin is critical but it is also non-trivial because it requires the network operator to acquire intimate technical knowledge of both modem performance and network performance. This technical knowledge must then be combined with knowledge of the business opportunity to engineer the noise margin for a given service capability. Excessive margin, and fewer customers will be reached, undermining the business case. Insufficient margin, and systems will be unreliable in service due to varying noise and cable conditions.
|
Written on 3:09 PM by ooe
Not surprisingly, moving DSL transmission technology forward beyond the state-of-the-art requires finding solutions to new problems. The technical challenges for VDSL are similar to those for ADSL – notably how to achieve high capacity transmission in the presence of high frequency noise and very steep losses in the twisted wire pairs which were originally intended only for low frequency voice telephony.
|
Written on 3:09 PM by ooe
The VDSL story begins around 1994 when it was realised that broadcast quality video could be delivered to a very high proportion of the UK population using the existing telephone network, without the disruption then being caused by cable digging. However, restrictions in BT's operating licence prevented it from offering broadcast entertainment using its network, and it was decided that an impressive demonstration of broadband capability was required to stimulate public debate.
Unfortunately, VDSL technology was not then commercially available and it was decided in the summer of 1995, to pull together a small highly skilled technical team at Adastral Park (formerly BT Laboratories) tasked with building an end-to-end VDSL demonstrator by the end of that year. An informal collaboration was established with DigiMedia Vision Ltd. (now a division of Tandberg), who loaned state-of-the-art MPEG-2 video coding and multiplexing equipment.
In less than four months a VDSL transceiver was designed and constructed from the ground up with general purpose components and using novel design techniques and training algorithms (now patented). In its very first test the modem operated at 23 Mbit/s on 500 m of cable for over 48 hours without a single error – an impressive achievement.
This early VDSL demonstrator delivered two simultaneous high quality digital TV programmes, plus high speed Internet over 600 m of typical telephony cable, and was sufficiently impressive for it to be shown to members of a House of Commons Select Committee early in 1996, and many other important influencers in the following months.
The prototype BT VDSL modem was based on the single-carrier modulation technique (SCM). The SCM variant selected was carrierless amplitude and phase (CAP-16) noted for its simplicity. A close collaboration with Nortel Networks' DSL team in Harlow was established who implemented a similar demonstrator using the competing, but more complex, discrete multi-tone modulation technique (DMT). This co-operation with Nortel provided valuable implementation insights for both SCM and DMT technologies, enabling BT to play an informed role in progressing VDSL standards.
This ground breaking work also provided the initial stimulus for the comprehensive programme of technical work carried forward since 1995 which underpins BT's acknowledged leadership in VDSL network technology and standardisation.
|
Written on 3:08 PM by ooe
The introduction of ADSL (Asymmetric Digital Subscriber Line) to drive forward the interactive broadband revolution is now well underway. ADSL is providing consumers with unprecedented high speed access to the Internet, and has sufficient capacity to deliver video and high speed data to the majority of homes. But ADSL is not the end of the story; demand for bandwidth will continue to grow and telecommunications network operators are developing the ability to deliver capacity far beyond ADSL using an exciting new technology called VDSL (very high speed digital subscriber line).
Like ADSL, VDSL operates on existing telephony wire pairs and is therefore a natural step for network operators who own an extensive telephony infrastructure. The key difference between ADSL and VDSL is that VDSL relies on deployment of optical fibre deep into the access network to reduce the distance over which data has to travel on the wire pairs, thereby reducing cable losses and increasing capacity.
The UK network includes both underground and overhead cables and is fairly typical of the networks to be found in most developed countries. This fact is relevant because it is important that VDSL technology is internationally standardised to reduce equipment costs, and it is helpful that most network operators have similar network constraints.
The distance from cabinet to customer is typically less than 1 km and provides the opportunity to access the final drop to connect VDSL transmission equipment. Fibre will used to backhaul the broadband services from the cabinet to the nearest broadband access node.
|
