6.5.2 Network use
We consider how the different types of network operator identified earlier will use these network types.
Internet Service Providers
ISPs are already introducing PC-Phone public telephony and PC-PC Internet named telephony using proprietary software. They wish to retain customers and gain revenue from voice. We expect that they will migrate from proprietary software to the use of SIP, however their solutions and technology will be heavily influenced by developments in Windows.
ISPs will be keen to foster the development of standard IP based telephones that use LANs connected to ADSL access, since this will give them the opportunity to provide telephony services to all users, not just those who are sitting at a PC. By not offering conventional telephony, they will hope to avoid the regulatory requirements that telcos have to meet.
Where they run their own networks, ISPs are likely to use IP based solutions that do not include ATM, which they regard as part of the telecom culture. They will however be keen to adopt techniques within their networks that can improve quality such as MPLS. MPLS will also enable them to offer more services to the business market such as virtual private networks.
ISPs will use the public Internet for connection to other networks but may augment it with techniques to improve quality for their voice traffic.
Internet Telephony Service Providers
ITSPs will take very much the same approach as ISPs.
Corporate network operators
Many operators who have extensive core networks provide advanced services including VPN to large corporates. These operators may also be Internet backbone providers. Examples are Worldcom and Cable and Wireless. These operators are likely to migrate their voice services onto IP within the next four years so that they can gain economies of scope and develop new services. Cable and Wireless for example has signed a contract for Nortel to replace all their circuit switches by IP based softswitches by 2004.
BTIgnite is developing a broadband network for multimedia and VPN services. This network is likely to compete to some extent with BT’s more traditional PSTN services.
These operators are likely to use the growing availability of ADSL access and subsequently of VoDSL on unbundled local loops to provide advanced services. Thus they are likely to move down market and provide VoIP services to SMEs and even some residential users from 2002 onwards.
Quality of service is likely to be a major concern for these operators. They are therefore likely to use MPLS within their networks. Some may send media packets directly on ATM without using IP.
Interconnection between these networks will initially be supported only using circuit switching call protocols such as ISUP because of the slow development of SIP standards to the level where they can support interoperability easily. We think that extensive interconnection for voice at the IP level using SIP will not occur until after 2005.
Telcos with leased infrastructure
These operators who do not have direct connections to customers have tended to use various technologies to minimise the costs of providing bypass services. They use a range of transmission and switching technologies. Some will move early to IP but a variety of solutions will remain. We expect that cost reduction and maximising the returns from existing investments will be the priority when the call based bypass market starts to reduce.
We think that telcos with their own infrastructure for direct connection to customers, eg BT and the cable companies, will follow two different and parallel approaches.
Where the voice service needed is still the PSTN, we do not expect there to be any strong incentive for migration from circuit switching to packet technology before 2005 at the earliest because these operators can continue to use the equipment that they have already purchased at marginal cost. There will however be two reasons why we think that they will want to introduce packet based telephony:
-
They want to gain experience of running a packet based PSTN type service
-
They need either to replace some of their oldest circuit switches or they need extra capacity in areas where traffic has grown rapidly or where there are new building developments.
For these reasons, we think that BT and the cable operators will introduce an overlay of packet based technology in very limited areas, starting from around 2004. This overlay will grow gradually according to need and some circuit switches may still be in use after 2010.
We think that low transmission delays will be the prime objective in the network design and therefore these operators may well choose ATM based solutions. In the case of BT, they may extend their ATM transit network with its BICC/ISUP call control down to the local level and not use IP at all.
In parallel starting from 2002, these operators will address the SME and the top of the residential market with broadband and VPN services delivered over ADSL and VoDSL in the same way as the corporate network operators. These services will include voice but initially may not have all the features required for PSTN substitution. They are likely to use SIP or H.323 for their call control.
6.6 Residential traffic forecasts
Traffic may be carried on IP for either of two reasons:
-
The caller uses IP eg
-
He makes a public telephony (E.164) call but from a PC using Internet access (currently called PC – phone)
-
He makes an Internet named telephony call from a PC to a person identified by “user@domain” (currently called PC-PC)
-
The network operator uses IP technology within its network
We consider the first of these cases as it will be the dominant one in the shorter term at the national level. We consider only residential calls because we have statistics for current residential call volumes but not for residential plus small business.
6.6.1 Qualitative
We think that there is still moderate potential growth for traditional type voice traffic as prices reduce, ie prices are still sufficiently high that there is some price elasticity. Some of this potential traffic, however, will be carried by mobile telecommunications and email.
We think that the current reduction in local and national call minutes of around 10% per annum is due to the growth of mobile communications which are substituting for fixed communications and to a lesser extent to the growth of email as a convenient means for sending short messages. Although the substitutionary effects of both will continue, there are some signs of saturation in the mobile market and so we think that these substitutionary factors will not stay at their current level.
We think that public telephony traffic from PCs will grow for two reasons:
-
it will compete on price and functionality with phone-phone, which is pure substitution
-
it will increase call minutes by reducing prices but this will only be a factor for 3-5 years for national calls. After that prices will be sufficiently low that prices will no longer influence calling times significantly
Internet named telephony from PCs will grow, probably much faster than public telephony, but initially with only a small proportion of the traffic being substitutionary and most being due to presence features and chat between small informal user groups. In the longer term Internet named telephony from PCs will have a significant substitutionary effect.
We have come across the following interesting instances:
-
The working non-technical wife of a telecommunications executive (who is well able to pay for normal telephony) comes from the North of England but lives in the South and prefers to use Internet named telephony from PCs to communicate regularly with her family in the North.
-
A group of 14-year old pupils at a private school in London use Internet named telephony from PCs in conference mode to do their homework together
Both cases are influenced much more by features than price.
Notwithstanding all the work on voice quality in the past in the ITU and ETSI, we expect that the threshold of acceptable voice quality is lower than the telecommunications industry believes.
We think that major enabling factors for growth in the use of PCs for telephony will be
-
the support of public and Internet named telephony in a future version of Windows, giving greatly improved awareness and ease of use. We think that this will start in 2002.
-
improvements in the quality achievable. It needs to become a little better to really take off but we think that this will occur by 2002-3.
-
flat rate charges for access, either FRIACO or ADSL. The “always on” feature of ADSL will help, but we expect ADSL to roll out only at moderate rate in the next two years and we think that the flat rate access charges are more significant than access speed as the speeds from dial-up can support IP telephony adequately
-
growth in integrated digital home systems including entertainment, but we expect that this will not be a factor until 2004 onwards because of copyright concerns.
Figure 37 shows qualitatively how we expect public and Internet named telephony from PCs to grow. The relative proportions of public and Internet named telephony are impossible to predict with confidence. Internet named telephony will grow more slowly initially as there is a square law effect based on the probability of both parties using PCs, but since most residential call minutes are traffic between regular correspondents (family and friends) the potential for growth is enormous.
Figure 37: Growth of voice traffic from PCs 
In the longer term, we expect growth in the use of standard IP based telephones, but this development will probably have relatively little impact on the residential market by 2005, so we ignore it in the quantification in the next section.
6.6.2 Quantitative projections to 2005
It is impossible to make reliable forecasts of the volume of IP telephony given the embryonic state of the market. We can, however, make projections based on reasonable and explicit assumptions.
The starting point for the projections is the Oftel statistics for end June 2000, giving:
-
The number of residential lines (24.5 million)
-
The number of residential public telephony call minutes from ordinary telephones and hence the number of residential call minutes per line per day. Calls to mobiles and calls to non-geographic numbers are omitted. In Q2 2000 this figure was 8.9 minutes/day for the sum of local, national and international, and 0.7 minutes/line/day for calls to mobile. For comparison, the average users of each residential fixed line together make some 7.4 minutes of calls per day from their mobiles, but this figure includes business calls .
We assume:
-
The number of residential lines for telephony purposes stays constant
-
The number of residential public telephony call minutes from ordinary telephones per line per day changes as follows from 2001 to 2005: -5%, -2.5%, 0% +2.5%, +5% giving a minimum in 2003. This pattern reflects the combination of fundamental growth in demand and mobile substitution. In Q2 2000, fixed terminal to fixed terminal traffic was reducing at 10% pa but we do not think that this reduction will continue
We show the estimate by National Statistics for end September 2000 of the number of homes with Internet access and our estimate of how this figure will grow, but we do not use any of these figures.
We use the Ovum29 estimate of the number of users of PC-phone services up to 2005 and assume one user per line.
We estimate that each line used for telephony from PCs will make 2.5 minutes of public telephony calls from a PC per day in 2000, but that this figure will grow at 12.5%.
Users of PCs will also be able to make Internet named telephony calls. We estimate that each such would like to make 6 minutes/day of Internet named telephony calls and that this will grow by 15%pa. This figure is larger than the public telephony figure because Internet named telephony will be more “chat” orientated. However such calls can only be made to other users of PCs for telephony creating a strong critical mass effect that in practice is likely to be rather less than a square law. The PC-PC critical mass factor takes account of this effect.
Figure 38 shows the results. The effect of substitution of PC based telephony for traditional public telephony has not been included (ie the figures for public telephony should be reduced by the substitution effect).
Figure 38: Projections of residential IP telephony minutes from PCs 
The forecast shows that public and Internet named telephony from PCs will each account for some 10% of the total residential traffic. It also shows the Internet named telephony growth overtaking the public telephony growth in about 2005. Figure 39 shows the results in graphical form. This figure also shows the Ovum forecasts30 that were arrived at by a different method.
Figure 39: Forecasts for telephony from PCs 
In terms of the rate of change, it is worth recalling that 10 years ago there was very little use of email, which began to take off around 1993, and very little use of the World Wide Web, which began to take off around 1994/5. Thus there is scope for substantial changes by 2005-2007.
6.7 Summary
Figure 40 draws the various qualitative forecasts together into a single table for the near, medium and long term future. The table illustrates the “jigsaw” nature of the developments.
Figure 40: Future scenarios Period | | Medium term
2006-2010 | Long term
2011 onwards | Services | Public telephony (E.164) universal Internet named telephony grows but only for informal groups | Public telephony (E.164) universal Internet named telephony becomes an any-any service | Both exist alongside each other | Service provision | Fragmentation | Consolidation and battle between ISPs and telcos | | Terminals | Analogue unaffected Growth phase for telephony from PCs | Analogue terminals start to decline Growth phase for standard IP telephones and integrated home systems | Integrated home systems | Access | Separate analogue and ADSL NTPs | Analogue access declines Growth phase for new IP based NTP | Standard IP based NTP | Networks | Growth of wholesale services Growth of global IP managed networks without interconnection | Replacement of circuit switched networks with SIP on IP or BICC on ATM. Media carried direct on ATM in many networks IP based interconnection implemented | Slow migration to all-IP as SIP gradually replaces BICC and IP is used without ATM and SDH |
7 Conclusion
The provision of voice related services over fixed telecommunications networks will change significantly in the next decade with increasing use of packet technology. The main conclusions about the way in which networks will develop are:
-
Public telephony will be provided increasingly over IP and ATM technology including the public Internet.
-
New Internet named telephony is already starting to grow as a means of communicating between informal groups. The presence feature where users are informed which members of their “buddy list” are on-line is likely to be very popular and lead to a new form of intermittent group communication. This service will eventually become an any-any service and exist in parallel with public telephony.
-
The early stages of the changes will be driven mainly by users who make calls from PCs initially for cost savings but increasingly for ease of use. Developments in the Windows operating system and new handsets with USB interfaces will be particularly significant.
-
Network operators will introduce packet based network technology. This will happed first for the support of new broadband services including but not dominated by voice. These services will roll out from 2002 and address the main demand for new features.
-
The replacement of existing PSTN services will happen much more slowly and not be complete even by the end of the decade because the existing circuit switches can continue to be operated at marginal cost. However there will be little new investment in circuit switches and so extra capacity will be provided by a packet based overlay, most probably based on ATM.
-
The choice of protocols for networks is not entirely clear. The long term preference seems to be SIP but much further development is needed to support interworking adequately and to provide the special features required by regulation.
-
Interconnection will depend on circuit switching until about 2005 when the standardisation has matured sufficiently for IP based interconnection of voice services.
-
Transmission delay will be a difficult and sensitive issue, and will have a major influence on network design. It may result in use of ATM continuing for a long time.
Contacts
John Horrocks
Horrocks Technology Limited
Bethany, Chapel Lane
Pirbright
Surrey GU24 0JZ
Tel: +44 1483 797807
Email: 100441.727@compuserve.com
David Lewin
Ovum Limited
Cardinal Tower
12 Farringdon Road
London EC1M 3HS
Tel: +44 20 7551 9213
Email: dmr@ovum.com
Share with your friends: |
