Voice over IP Equipment andService Markets

Elizabeth McPhillipsInternet Research InstituteHP Laboratories BristolHPL-IRI-1999-00527th August, 1999*

VoIP, internettelephony, IPtelephony

Internet telephony, Voice over IP, has attracted much attentionrecently due to its potential to significantly reduce long-distance and international voice communications costs, andbecause it presents entirely new and enhanced ways ofcommunicating. This report introduces the main topicsconcerning the VoIP market - regulation, technologies, servicesand players. The growth and segmentation of the market is alsodiscussed, and statistics regarding the future are given.

∗ Internal Accession Date Only

Copyright Hewlett-Packard Company 1999

1

Voice over IP Equipment and Service Markets

Elizabeth McPhillipselimcp@hplb.hpl.hp.com

July 1999

1. Introduction to VoIPVoice over Internet Protocol (VoIP) hasattracted much attention recently for anumber of reasons, the most notable beingthat it has the potential to significantlyreduce the cost of long-distance andinternational voice communications, andthat it introduces entirely new andenhanced ways of communicating.

The fundamental applications of VoIP arevoice, fax, data and multimedia.Traditionally ‘voice’ has suggested voicetelecommunications alone, however ‘Voiceover IP’ is a general term used for all areasof traditional telephony – and furtherapplications – that are to be offered overthe public Internet and over private IPnetworks. As a result, fax, multimedia,and real-time data services can beconsidered sub-segments of VoIP.

VoIP had a difficult start, partly due totechnological limitations, and partly due tothe slow acceptance of new technologiescharacterised as being of poor quality andreliability. Technological improvementsand strong public demand for lower phonetariffs are resulting in the widespreadmarket acceptance of VoIP.

VoIP services are fast becomingpracticable alternatives to traditionaltelephony services. Government regulatedPSTN (Public Switched TelephoneNetwork) services, with artificially highprices, are left vulnerable as this newtechnology steps in. Quality and otherchallenges to VoIP are being addressed,and the worldwide boom in Internet usagemakes it a conceivable substitute, possiblyresulting in the long-predicted integrationof voice and data networks.

Current applications for VoIP technologyare primarily focused around alternatelong-distance voice services. Incumbentservice providers are recognising the VoIPopportunity, as are non-traditional serviceproviders and next generation telcos.

Voice over packet had been researchedduring the 1970’s and 80’s, but realdevelopments in VoIP did not begin until1995. Israeli firm VocalTec initiated theIP telephony market in 1995 with softwarethat allowed a voice connection betweentwo PCs over an IP-based network.Several other packages began to emerge,and in 1997 Delta Three launched the firstphone-to-phone service for commercial use.

Although there are current market nichesand segments where VoIP prevails, it hasmajor obstacles to overcome before itbecomes a mass-market success. Theseinclude regulation, quality of service,standardisation, infrastructure and networkmanagement, and bandwidth.

Section 2 of this report discusses theregulatory environment and how this mayaffect the growth of VoIP. Section 3briefly introduces the methods by whichVoIP may be implemented. Section 4looks at the necessary equipment and itsmarket. VoIP services, both network andretail, including value-added services, areoutlined in Section 5. The advantages ofand challenges to VoIP are discussed inSection 6. Independent ITSPs andtelephone companies, and their VoIPstrategies are examined in Sections 7 & 8.Section 9 looks at the VoIP marketsegmentation and uptake, and Section 10gives statistical forecasts for the future ofthe market. Section 11 wraps upsignificant market observations. A brief

2

glossary is also provided, with entriesmarked [glos].

2. Regulation of VoIP

Regulation of VoIP is largely questionable.Telephony services have traditionally beenheavily regulated. Most governmentsoperate telecommunications monopolies,and deregulation has come recently inothers (1996 in the US, 1998 in most of theEU).

Most governments’ policy (both indeveloped and developing countries)towards the Internet has been to encouragegrowth and competition by adopting alaissez-faire attitude. The rapiddevelopment of the Internet has, in large,been due to this lack of regulatoryintervention. The introduction of mostforms of regulation would almost certainlyretard innovation in the Internet, and wouldcreate a barrier to entry for start-upcompanies. Those that would stand to gainfrom this would be, in general, operatorsthat have enjoyed a monopoly position formany years. Both the FCC (FederalCommunications Commission) and theEuropean Commission welcome this aid toforcing down international phone tariffs.Regulatory constraints that are specificallyrelated to the Internet may put new entrantsat a severe disadvantage, and would assistthe continued over-pricing of long-distanceand international telephony. However,general telecoms regulation may haveadvantages for the small VoIP provider.

The main barrier to changing telephonecompanies, for a consumer, is having tochange phone numbers, and the associatedinconvenience. Number portability allowscustomers to keep their number whenchanging companies. When phone-to-phone VoIP becomes convenient andcommonplace, the ability to offer a totalVoIP service without having to changephone numbers will work to the advantageof smaller providers.

With the introduction of high-speed localloop [glos] technologies, the ability toaccess the local loop will become necessaryin order to provide a competitive service.Regulation has resulted in the unbundlingof the local loop in the US, Hong Kong andsome European countries. Recently (July1999) BT was ordered to open up theirlocal loop. This will give the opportunityof high-speed access provision to smallproviders.

Although there is enormous growthpotential in the short term, with increasedchoice and competition, the uncertaintyover the future of the regulatory structureis still impeding investment anddevelopment. Telcos, not wishing toobliterate existing profitable business, maybe banking on regulation, assuming thethreat of VoIP will peter out as ITSPs(Internet Telephony Service Providers)become classified as voice telephonyproviders. Many others will not want toinvest in VoIP solutions until its technicalproblems are overcome.

Enhanced data services are those that canbe provided economically by independentservice providers (non facilities-basedservice providers, [glos]) in competitionwith network providers, as long as they cangain access to the networks. Examples ofenhanced services are the provision ofcontent for premium rate services and theprovision of retail Internet services. ITSPs,and ISPs (Internet Service Providers) ingeneral, as ESPs (Enhanced ServiceProviders), are not subject to long-distanceaccess charges. Telcos are urging thisregulatory position to change. As ITSPsbecome increasingly competitive withtelcos, it is argued that they should beclassed identically. At the same time, it isvery difficult to distinguish betweendifferent types of Internet traffic, and hencea method of levying access charges onITSPs seems almost impossible.

The European Commission has definedvoice telephony as ‘the commercialprovision for the public of the direct

3

transport and switching of speech in realtime between public switched networktermination points, enabling any user to useequipment connected to such a networktermination point in order to communicatewith another termination point’ [1].Thus the following conditions must be met:• the service offered must be part of a

commercial offer• the service is provided to each member

of the public• the service involves direct transport and

speech in real time• the service is provided between public

switched network termination points inthe fixed telephony network

VoIP clearly lacks some of these criteria,most notably real-time transmission. Italso falls short of traditional telephony inthat it is not fully reliable, ubiquitous, nortransparent.

VoIP is not a generally available voicetransmission service, but an applicationavailable to those who have alreadysubscribed to a basic Internet package.Because of this ‘closed-user group’offering, the European Commission maynot force ITSPs to pay access charges.Also, due to the telephony restrictions inEurope, VoIP is seen as an effective way ofintroducing competition.

It has been suggested [2] that additionalregulatory movements may affect the VoIPindustry. These include market entryrequirements and entry rules:

Because most Internet services have beenclassified as non-basic or non-voiceservices, providers have not been subject torestrictive licensing and market entryrequirements applicable to othertelecommunications providers. As VoIPbecomes a serious competitor to traditionaltelephony, regulators may start to imposeregistration, licensing, or otherrequirements on ITSPs. These processescan be time-consuming, increasing thetime-to-market of new entrants.

If ITSPs are classified as carriers of basicvoice services, they may be required tocomply with routing restrictions and rules.These may restrict services that bypassaccounting systems. However, given thecomplex topology of the Internet, it wouldbe virtually impossible to monitorindividual packets and impose routing ruleson them.

3. Methods of VoIP

Initially VoIP was easily dismissed. Clientsoftware gave poor quality, similar to CBradio, and allowed only one party to speakat a time. Both communicating parties hadto be online together and had to havecompatible software running on their PCs.However, potential for the service was seenby vendors such as Netscape, Intel andMicrosoft, all launching VoIP products in1996. By this stage, both computingpower and Internet backbone capacity hadimproved such that much better voicequality was available with VoIP. At thesame time, VoIP communications receivedITU (International TelecommunicationsUnion) standardisation and by 1997standards-based products becameavailable.

The basic necessity of all early VoIPproducts was that the users at each end hadan Internet connection. The developmentof the VoIP gateway has changed this suchthat PC-to-phone, and even phone-to-phonecalls can be made. This has removed oneof the most important barriers to broadmarket acceptance. With gatewaysinstalled, users need not have PCs, specialsoftware or even an Internet connection toplace calls that are routed mainly over thepublic Internet. Gateways allow users tocircumvent the PSTN and its tariffs, usingthe Internet for voice communications ofvariable quality, and for significantlyreduced prices.

4

Many new operators* and equipmentmanufacturers/vendors have emerged sincethe release of the Internet Phone product,the first robust and scalable VoIP productand the initial VoIP service offered byDelta Three in late 1995. Client softwareis now supported by most operatingsystems (Windows, Unix, Mac).

PC-to-PC telephony is the original form ofVoIP. Both users are required to be onlinebefore any connection is set up, and mustuse compatible software and multimediacomputers. It is not practical to separatelytariff or regulate this form of VoIP, as it isdifficult, and probably counter-productive,to try to distinguish audio bits from otherbits. It is also difficult to distinguishbetween real-time, full-duplexcommunication and time-insensitive store-and-forward information. Unless theInternet is regulated as a whole, it is notfeasible to regulate this form of VoIP.

PC-to-phone telephony arose as IP-to-PSTN (and PSTN-to-IP) gateways becameavailable. A gateway packetises andcompresses voice traffic from the PSTNand places it on an IP network, andassembles and decompresses traffic in theother direction.Regulating PC-to-phone VoIP is adifficulty. The originator of the call paysno access charge, as to the service providerthis looks the same as a PC-to-PC call.This greatly reduces access charges as theoriginator usually pays about two-thirds ofthe overall access charge. At theterminating end, the service provider mayprovide termination services to both localand long-distance callers, but cannot knowwhere any particular VoIP call originated,spurring problems with regulation andtariffing. Similarly, with phone-to-PCVoIP, there is no access charge at thetermination point, as the originator cannotknow where the call will be terminated.

Phone-to-phone VoIP eliminates the needfor a PC altogether, using the Internet to * see http://www.iptelephony.org/frame/providers.html

carry the voice between two gateways.Phone-to-phone VoIP calls can now offergreater quality than PC-to-PC or PC-to-phone VoIP calls, as the former may beoffered over a dedicated IP network. Thereis breakout to the PSTN at both ends, butbecause the originator cannot know thetermination of the call, and vice versa,problems regarding regulation and accessfees will exist.

Fax over IP (FoIP) is arguably superior totraditional fax because it does not requirereal-time transmission. Transmitting a fax,which has traditionally been associatedwith the PSTN, is no different fromsending any other file. Fax messages areable to absorb delay and retransmittedpackets without any negative impact on thefinal received message.

Multimedia conferencing will involve thecombination of real-time voice, video anddata transmission over the Internet.Although standards are emerging, it willlag behind other VoIP applicationssomewhat due to the substantially greaterbandwidth required for multimediaapplications.

4. VoIP Equipment

In order to provide VoIP solutions, theInternet and other IP networks mustinterface seamlessly with the PSTNthrough gateways. These gateways will beone of the key areas of revenue for VoIPequipment vendors. IP networks will alsohave to acknowledge the resourcereservation and prioritisation schemesneeded to guarantee quality of service fordelay-dependent data.

The hardware required to provide VoIPservices ranges from standard networkingequipment such as routers and switches, togateways, to IP phones.

4.1.1 IP PhonesIP phones function as regular phones do,but instead of connecting to the PSTN,

5

they connect to an IP data network andtransmit voice messages through IPpackets.

Presently, IP phones are very expensive,and using VoIP can be cumbersome andinconvenient. Traditional telephones arecheap to buy and simple to use, andmarginal costs are close to zero, allowingprices to be reduced hugely. Hence IPphones (and VoIP in general) have todevelop further in the way of ease of use,and decrease in price, before they caninvade traditional telephony space.

Until now, IP telephony has been difficultfor the average user to adapt to because theinterfaces were inconvenient. By bringingmany of the familiar and necessary featuresof traditional telephony to the IPenvironment, the move towards seamlessintegration of the PSTN and IP networkshas begun.

Selsius, purchased by Cisco in late 1998,has introduced a phone that provides all ofthe functionality found in a traditionalphone, and connects directly to an IPnetwork. This signals an important shift inthe telephone equipment industry, possiblysuggesting a new opportunity for PBX(Private Branch eXchange, [glos]) andtelecoms service vendors. In theory, IPtelephones are as functional as traditionalphones, whilst being able to add newcapabilities.

Nokia’s IPCourier Ethernet Phone providesPBX functionality (such as multiple lineappearances and call control features)without a PBX. It is an Ethernet telephonewith a familiar interface that links directlyto an IP network. It also supportstelephony features such as call waiting,caller ID, call forwarding, and calltransfer.Nokia’s IPShuttle product is a customerpremises device that allows standardtelephones to be plugged into an IPnetwork. It is aimed at residentialenvironments and is designed specificallyto serve new voice and data services being

offered by cable companies and telephonecompanies [3].

Siemens has also released (April 1999) anIP telephone with enhanced capabilities [4]- a new interface, ability to retrieve thenames and addresses of IP callers, andinformation regarding unanswered calls.Their HiNet LP 5100 IP telephone is tocost $425, but this price will fall rapidly.

Ericsson has released the telephoneDoubler, which allows use of a virtualtelephone line, enabling the user to receivephone calls (over IP) without having to logoff the Internet. The voice traffic is carriedover IP between the user’s PC and thegateway, with the rest of the call carried onthe PSTN, thus Quality of Service (QoS,[glos]) is directly dependent on the user’sconnection speed and on the traffic theygenerate.

4.1.2 VoIP GatewaysGateways act as interfaces between thePSTN and IP networks. When making aVoIP call, there is usually breakout to thePSTN at each end, for the first and lastlegs of the link, with the Internet or anotherIP network used for the trunk connection.

The gateway accepts connections from atraditional phone or fax machine,determines the termination point of the call,and decides what is the best and cheapestway to route the call (i.e. how much of thelink will be over the PSTN).

Although gateways were originally aimedat business customers, they areincreasingly being aimed at ISPs andcarriers wanting to offer VoIP services as acompetitive advantage. As scalabilityincreases, the opportunity is there to supplyproducts to service providers rather thansmaller enterprises.

Gateways tend not to interoperate well,leaving users confined to the network oftheir provider, or one formed throughpartnerships. H.323 [glos] is likely tobecome the standard interoperability

6

protocol, and this standardisation willhappen soon (by the end of next year, 2000[5]) due to pressure from service providers.

Motorola has teamed with Vsys (June1999) to develop a VoIP gateway based onits Vswitch VoIP solution for intelligentnetwork applications [6]. Vswitch is aUNIX-based software product thatprovides interfaces for both H.323 and SS7(Signalling System no. 7, [glos]) standardsto support voice, data and fax. Itscapabilities include billing andprovisioning, and advanced intelligentnetwork services such as freephonenumbers and single-stage dialling. Theproduct is to be commercially availablelater this year.

Gateways can be implemented through thefollowing methods:• router-based voice modules• concentrator-based voice modules• PC-based servers

PC-based gateways move data through aninterface, across the LAN and to the router.These standalone devices cannot movedelay-dependent traffic to the front of therouter queue, however they have severaladvantages:• they are independent of the router and

can be installed in mixed equipmentnetworks

• they are easily scalable• they do not require router upgrade• their market is more competitive than

other gateway markets, resulting inlower prices

• they were specifically designed for thispurpose

Router- and concentrator-based gateways[glos] can reduce delay time as they caneliminate the extra hop PC-based gatewaysrequire. This can improve the quality ofthe system. The traffic is moved directlyinto the router queue and to the WANinterface. Simple forms of prioritisationmay also be employed.

4.2 Equipment Market

Suppliers to the VoIP market must bridgethe dynamic world of the Internet and therather more static world of traditionaltelecommunications.

The VoIP equipment industry grew fromsmall players such as VocalTec. As theindustry has grown, much acquisition hasoccurred. Nortel acquired Micom, thegateway vendor, and Bay Networks, Ciscoacquired Selsius and has incorporatedvoice modules into its routers, and Nokiahas acquired Ipsilon, an IP switching firm.Smaller vendors are teaming with largerplayers*, such as software provider eFusionannouncing deals with Ascend (who havemerged with Lucent) and 3Com, providingIP gateway functionality through DSPs(Digital Signal Processors, [glos]) in theformers’ equipment. These partnershipsintend to build upon the experiences ofeach member. This will lead to newproducts as well as the adaptation ofexisting technology.

Although the market is growing, Internetequipment is not recognised as being asreliable as PSTN equipment. However, asthe world of telecommunications movestowards IP, new equipment will be in theform of data equipment. Hence dataequipment will be updated to handle voicerather than vice-versa. This will give anadvantage to data networking equipmentvendors, as this is where they haveexperience.

Telecoms operators with their own accessnetworks have been the main customer oftraditional suppliers such as Nortel, Alcateland Siemens. There is often a supplierrelationship that has been built up overyears with the main provider in theircountry.Data networking suppliers understand IPnetworking better than traditionalsuppliers, but may not satisfy the qualityand management requirements of acompany as strong as a telco. Thesecompanies have experienced huge growth

* see http://www.iptelephony.org/GIP/partners/

7

in recent times, and need new applicationsto continue this growth. VoIP is a greatmarket opportunity for them. As well aswith the tremendous number of newentrants, these suppliers are developingrelationships with traditional telcoequipment suppliers, and either byacquisition, alliance or marketing aremaking relationships with the telcosthemselves.

Large data and VoIP vendors have greatexperience with packet switching and haverelationships with ISPs and ITSPs. Theyneed to build on their expertise in real-timecommunications. The key challenge forthem (and for other businesses operating inthis area) will be to understand theintegration of voice and data, and to beable to realise this integration bothtechnologically and economically.

Most data networking vendors have addedVoIP capabilities to their equipment, suchas routers and remote access equipment.They are also extending their networks toenable more sophisticated functions. Thisis the strategy for the integration of voiceand data networking in order to create amulti-service network.

From a provider perspective, deployingVoIP gateways is much more expensivethan deploying PSTN switching equipmenton a per-port basis. VoIP gateways costan average of over $1,800 per port in 1997and now (1999) have fallen as low as $600per port. The same capacity (i.e. per line)for traditional telephony costs about $100-$200. Carriers are placing enormouspressure on vendors to lower costs to the$100-per-port range within the next year.Increasing scale (more ports per product)and economies of scale will lower prices.Standardisation will add to economies ofscale, and in all high-tech industriesengineering tends to improve, thus loweringprices.

Although the Internet enjoys significantlymore users in the US than in Europe, themarket for VoIP equipment is likely to be

split approximately equally between thetwo geographic areas. Because telephonyprices in the US are much lower than inEurope, the opportunities for new entrantsare lower in the US, and hence theenthusiasm is generally reduced. However,due to new IP networks being built byITSPs, and the fact that the US is a moreIP-developed region, there is still expectedto be a high uptake of VoIP.

The worldwide market for VoIP gatewayswill reach $1 billion by the end of nextyear, growing by 75% the following year,and a further 40% in 2002 [5]. Thesuggested decline in rate of increase couldbe due to the falling prices for suchequipment. The number of ports shippedwill also increase greatly over the nextcouple of years, to over 8 million by 2002,but rate of increase will decreasesignificantly after 2000.

The uptake of VoIP may also affect theWAN switch carrier market, with switchespurchased specifically for VoIP purposesto increase from less than 1% now toaround 6% by 2002, producing revenue ofnearly $.25 billion [5].

5. VoIP Services

The VoIP industry is changing from oneessentially catering for hobbyists into onein which services are offered to the generalpublic. These services are low in qualitybut tariffs are substantially lower than forPSTN telephony. The next phase of thistransition will be to a market where qualityis comparable to the PSTN andfunctionality is significantly moreadvanced.

Service providers offering reduced priceVoIP services can do so due to theefficiency of packet switching over circuitswitching, and due to the lack ofregulation, and hence access charges.Traditional telcos will reduce their PSTNprices, develop their own VoIP strategy, orface the loss of substantial market share.

8

Although many facilities-based operators[glos] have their primary focus on thewholesale capacity market, almost all ofthem are offering end-user consumerservices. In addition to this, there arenumerous ISPs and ITSPs offering VoIPservices, with many new entrants makingtheir presence known in the market.

The key focuses of retail VoIP providersare the long-distance and internationalmarkets. Due to the avoidance of accesscharges (long-distance and international),tariffs can be significantly cheaper forthese markets. International markets havegenerally been difficult to enter, butsmaller providers can now use the Internetto offer international services. Aside fromdifficulties obtaining access to localnetworks, it is also the case that in the localloop, the trunk link length is reducedsignificantly (and sometimes eveneliminated completely), and because VoIPtends to use the PSTN for first and last legconnections, using the Internet for the shorttrunk length would reduce the cost benefits.In the longer term, local access will becomemore important as last mile technologiesbecome increasingly sophisticated.

Fax will also be of importance, and willmove to IP before voice telephony due toits latency tolerance. Sending a fax overan IP network is no different from sendingany other file, so much fax traffic will usethe Internet as its trunk connection. Anestimated 30-40% of long-distance andinternational traffic is fax [estimate basedon numerous sources], and this can be veryeasily switched onto the Internet. Fax overIP is discussed below (Section 5.4).

As an increasing number of providers areinterested in providing long-distanceservices, those that own the networks are ina prime position to take advantage of this.

5.1 Network ServicesThe communications network can bebroken down into two sections - thebackbone network and the access network.

Backbone network providers supplywholesale bandwidth to service providerscloser to the end user. They may own thenetwork or lease it from other backboneproviders. Backbone providers tend to beincumbent telcos or large traditionaloperators, but a new generation has begunto emerge, including companies such asLevel 3 and Qwest, building brand new IP-based networks. Large ISPs and cableoperators are also involved in offeringbackbone capacity.

Technologies such as ATM (AsynchronousTransfer Mode, [glos]) and frame relaywill allow the backbone networks to carrydifferent types of traffic, and to offerquality of service guarantees for real-timeservices. In situations where quality andlatency are the issues, it is likely that IPover ATM will be the dominanttechnology, however where efficiency isimportant, IP may be run directly overSONET (Synchronous Optical Network)(or SDH, Synchronous Digital Hierarchy)[glos, and see Appendix 1]. Whatever thephysical/link layer technology, the futurebackbone appears to be IP - it is auniversally accepted network protocol andis independent of lower layer technologies.

QoS issues are the responsibility of thebackbone operators, as are Virtual PrivateNetwork (VPN, see Section 5.3)obligations offered as value-added services.

Access networks essentially belong toincumbent telcos and, to some degree,cable operators. It can be difficult to getaccess to the local loop, and theconstruction of entirely new accessnetworks is economically impractical.Although regulation in some countries ispushing for an unbundled local loop, itsowners still have enormous marketstrength.

The local loop speed will increase greatlywith the mass deployment of high-speedtechnologies such as xDSL (DigitalSubscriber Line), cable, satellite and radio.

9

The improvement of VoIP qualitycompared to that of the PSTN may have animpact on the local loop and its owners.This will be discussed in Section 8,Telephone Companies and VoIP.

Unlike telcos, most small providers havenot evolved sophisticated billing systems,and it can be difficult for them to developthese systems over short periods of time.This has pushed the idea of pre-paid callingcards by small providers. This offers manyadvantages to providers and customers.Customer identification and authenticationis simple – the customer dials in hisaccount number and his information iscaptured. If the calling cards are pre-paid,no bad debts can accumulate. Essentiallyno billing is involved. The cards can beobtained and recharged directly through thecompany’s web-site with automatic billingby credit card. A manager havingemployees with frequent changes to theircalling requirements can do this. Althoughthis method gives a good short-term way ofpromoting the service, in the longer term,users will require more convenience.

5.2 Retail ServicesThe enhancements to traditional telephonyservices will be the main market attractiononce the tariff arbitrage has dissolvedsomewhat, as it appears it will.

Current PSTN caller ID allows thereceiving party to see the number of thecalling party and nothing more. An IPenvironment can support more information– who the caller is, where he is, the requestfor a multimedia connection, or any otherinformation he wishes to send.

The call waiting function available withVoIP is greatly enhanced compared to thePSTN version. When a call arrives, a newwindow can be displayed on the user’sscreen showing the caller’s name andnumber (and any other information he hassent). Then the user can choose how tohandle the call. The call can be answeredon the phone (either by disconnecting theInternet connection and taking it on the

PSTN, or by receiving it by VoIP over theInternet connection). The call can beforwarded to another number, such as amobile. The call can be forwarded directlyto voice mail or a busy tone can beprovided to the caller (these apply if theuser does not wish to be disturbed). Amessage can also be played for the caller tohear, instructing him what to do next.

Click to talk - an individual can initiate orestablish a voice connection by clicking ona web page. This service has applicationsfor catalogues and call centres, amongstothers, enabling a customer to speak to arepresentative while looking at acompany’s web site. The customer canclick on an item to initiate a call to thecatalogue and speak to the salesrepresentative through their PC withouthaving to take down the Internetconnection. This is discussed further inSection 9.1.1, Business Usage.

The integration of services such as voicemail, fax, email and other information willbecome important as customers becomemore mobile. A number of mobiletelecommunications companies havealready begun to offer some sort ofintegrated service, allowing customers tocheck their email over their mobile phones.These services may also allow the sendingof messages from the Internet directly to amobile phone. Deutsche Telekom offers anextension to this, by way of their NOVICEsystem, using voice-synthesis software toconvert email messages into speech, andcustomers can have their messages readover the phone to them. Other companiesprovide services whereby information suchas football results or stock market quotescan be obtained on mobile terminals viaSMS (Short Message Service, [glos]) [7].

Real-time administration (billing andprovisioning) is becoming a service initself. The unification of bills for allservices, and the ability to view bills online(real-time, up-to-date) will be seen as ahuge convenience. Provisioning of servicesmay also be available ‘on the fly’ over the

10

Internet, e.g. do I want call waiting servicesfor this session? Yes, click, and they areavailable.

As Internet penetration grows, the technicalliteracy of the average user will decrease.Thus, a more robust and comprehensivecustomer service will be necessary to serveless savvy users.

5.3 Virtual Private NetworksVPNs are private data networks that makeuse of the Internet, maintaining privacythrough the use of a tunnelling protocol andsecurity procedures. They can becontrasted with a system of lines leasedfrom a telco that can only be used by onecompany and which can be incrediblyexpensive. A VPN gives a company thesame capabilities at a much lower cost byusing the public infrastructure rather than aprivate one.

Because market demand has not yet beenhigh, Internet VPN services have not beenpushed as a key element of most providers’offerings. It is supposed that the marketwill be demand-lead, and so although manyproviders have the technical ability to offerthese services, they have not seen thenecessity to promote it. However, they dofeel that VPNs will change the serviceprovision market dramatically and offergreat opportunities to providers.

Consumer and small business VPN VoIPservices over the Internet are likely to beoffered by non-facilities based carriers,whereas facilities-based carriers aretargeting large business customers.

Because of the individuality of eachnetwork, services will be priced accordingto the specific requirements of thecustomer. No particular markets havebeen targeted yet, with providers offeringservices to individual organisations.However, certain market segments presentgreat opportunities for early adoption,including the IT sector, firms with a largenumber of small sites such as travelagencies, estate agents and insurance

brokers, and high-tech manufacturing firms[5]. Targeted companies will also bedetermined by size band, large companiesbeing a prime target for long-termcontracts, and medium-sized firms showingshort-term potential.

5.4 Fax Over IPIn spite of predictions made in the face ofemail, the fax market is growing, notwaning. Despite the popularity of emailfor general communications and filetransfer, some functions, such as sending ahand-written, signed document cannot beaccomplished effortlessly. Speed andreliability of email can also beunsatisfactory.

Traditional fax machines are connecteddirectly to the telephone network, just as atraditional telephone is. Phone numbers letvarious local and long distance carriersinitiate connections and deliver faxes.Internet fax servers collect traffic fromstandalone fax machines and route it to theInternet for delivery. The user dials thedestination phone number, but it is theInternet fax server that interprets thedialled digits, not the PSTN, and the fax istransmitted via the Internet to the remotefax server.

The global fax market has been able toflourish due to the adoption of fairlyconsistent standards that are usedworldwide. One of the problems of usingfax over the Internet has been the lack ofstandardisation. It seems likely that thefirst standards-based FoIP will be a mail-type standard to transport fax image files,but session-based and real-time fax shouldfollow closely as more immediatedocument delivery and confirmation ofdelivery become necessary.

Already an extensive number of faxservices are available over the Internet,mainly by large backbone operators(AT&T, MCI, Sprint) or companiesfocused on providing FoIP (FaxSav,FAXiNET).

11

Mailbox services are provided, comprisinga fax or unified messaging mailbox thatcan store and forward a user’s faxes, canbe configured to automatically forwardmessages to another location, and can oftenallow users to check voice, fax and emailsimultaneously.

Fax multicasting can be provided, allowingthe sending of a single fax to multipledestinations, be they fax machines or PCs.

Overflow services take into account thatthe receiver’s fax may be busy or notanswering. The fax is stored by the serviceprovider and then automatically forwardedto the destination machine at a later time.

Interactive fax/fax-on-demand servicesallow a customer to store documents on theservice provider’s server, such that clientswithout Internet access may retrieve thesedocuments by calling into an interactivevoice response system and requesting them.

Message translation can convert faxmessages to email, email to voice, or othertransformations for the delivery ofmessages. This is becoming a significantfeature of the unified mailbox.

The primary target markets for FoIP (andunified messaging) will be the SOHOmarket and mobile professionals. Theformer will have increased functionality,whilst saving money, and the latter canreceive all messages – voice mail, emailand fax – at once.

The FoIP market will initially be mainly aninternational one, due to the pricedifferentials between using the Internet andthe PSTN. Until 1997, virtually all of theglobal international fax market was PSTN-based. According to Probe Research, by2001, FoIP will account for 10% oforiginating international fax, and by 2006,almost 30%.

Multinational companies may be able tosave a substantial part of their average of$15m a year on fax-related transmission

charges (based on Fortune 500 fax usage)[8]. Revenues lost by PSTN operatorsmay be in the region of $1bn by 2002, andmay reach $2bn by 2006 [9].

6. Advantages of and Challenges toVoIP

6.1 Advantages of VoIPThe ISP market today is driven largely bya tariff arbitrage. Although QoS isunsatisfactory compared with traditionaltelephony, people wishing to make long-distance or international calls over theInternet may do so cheaply, saving 20-100% on a call by bypassing the PSTN[9].

There are three main reasons for the priceadvantage [10]:1. Regulators have ruled that VoIPproviders are not subject to the long-distance access charges levied ontraditional providers2. VoIP providers avoid the payment ofsettlement charges to foreign operatorsterminating their calls (of which USoperators pay approximately $5bn peryear) [11]3. The networks used are more efficientdue to compression and packet switching.

The technical and economic advantages ofpacket switching are numerous. It appearsthat VoIP has a long-term cost advantageover traditional telephony based onefficiency and lower equipment costs.

Packet-switched technology uses theinfrastructure with more efficiency than thecircuit-switched technology used intraditional telephony. Although the cost oftelephony switching equipment has fallensharply in price, so has the cost ofcomputer equipment. Port prices for VoIPare relatively high but are falling rapidlyand will soon undercut prices per line fortraditional telephony. Low cost routers arebeginning to replace switches.

12

The unit economics, although debatable,seem to favour packet switching, withestimates suggesting that packet switchingis almost four times as cheap as circuitswitching at the per byte level [12].

The data overhead associated with packetswitching is far less significant than thecapacity wasted in circuit switching whenno information is being conveyed.

Much swifter compression update can bemade on the Internet than on the hardwareof the PSTN. On a circuit-switchednetwork, all hardware in the network mustbe updated in order to progress withadvances in compression, whereas anInternet client on a standard PC canimplement the latest technologies withoutconcern for the rest of the network.

Multimedia services on IP networks willallow one network to be used for telephonyand data, as well as more sophisticatedmultimedia services. Although there is notyet any practical evidence to support theargument that managing one network ismore economical than managing two, theargument is straightforward.

Even in an unencrypted form, speechcarried over the Internet is considerablyharder to wiretap than analogue speechcarried over a copper wire pair. Whereencryption is used, decryption by anunauthorised party would be immenselytime consuming, taking many weeks ormonths for a supercomputer to decrypt atwo-minute phone conversation [12].

VoIP has not, until now, been a convenienttelephony mechanism, being cumbersomeand requiring awkward dialling, wherebythe user is required to first dial an accessnumber, wait until this call is routed to theaccess IP telephony switch, and only thencan the destination number be dialled.Providers have begun (early 1999) tointroduce VoIP services that will no longerrequire the user to dial any differently thanhe would when making a regular call [13].Korea Telecom has been one of the first

providers to integrate the PSTN and IPnetworks in this way [14]. This is animportant development to ensure thespeedy evolution of VoIP. LucentTechnologies [15] and Nortel Networks[16] have developed packages that willsupport the delivery of enhanced VoIPservices, including the above, single-stagedialling.

Once feature parity exists, future uptake ofVoIP will depend largely on theintroduction of enhanced applications. It isassumed that by this stage the pricedifferential will be much less significant,due to lowering PSTN prices and/orregulation.

Applications already available include callwaiting, caller ID, multicast, andmultimedia conferencing. Simpletraditional features can be much richer overa packet-switched network. The benefits ofsuch enhancements are obvious. The needfor separate lines for voice and datadisappears, and flexibility and functionalityare increased.

The possibilities for such applications onIP networks certainly extend beyond thatavailable on the PSTN, and to applicationsnot yet envisaged.

VoIP is driving the public network towardsan open, client-server topology.Proprietary switching systems and legacynetworks have retarded integration of voiceand data for many years, but competitionin the area of telephony has initiated themove towards an open network. Thiscould be the trigger that brings about trueconvergence of voice and data networks,but this cannot happen until cost parityexists between packet-switched voice andcircuit-switched voice technologies [17].VoIP equipment, although falling rapidly inprice, is still considerably more expensivethan traditional equipment. This willchange, but will take a number of years todo so [18].

13

The ubiquity of IP networks in enterprisesis ensuring a large addressable market forVoIP equipment as well as facilitating theintegration of VoIP equipment into existingnetworks. Enterprises are beginning to putVoIP on their WANs and LANs, cuttingcosts as the voice is treated as data. Onsuch private networks, control of data ismuch easier than on the public Internet,with QoS exceeding that of VoIP on theInternet.

The value of any telecommunicationsnetwork increases exponentially accordingto the number of users/computers attachedto it (Metcalfe’s law). As more and moreIP equipment is attached to the Internet, thebenefits derived from being part of a VoIPnetwork will increase, fostering furtherdemand.

To sum up, there is huge opportunity in theprovision of VoIP. Presently, the mainattraction is the low tariffing that can beoffered by bypassing various accesscharges traditional providers must pay.VoIP is no longer a technology employedmerely by hobbyists. A new generation ofusers has emerged, and as convenience ofuse and richness of applications becomesatisfactory and attractive, VoIP willbecome increasingly popular amonggeneral telephony users.

6.2 Challenges to VoIPThe low quality of service and unreliabilityof the Internet are the main marketrestraints for VoIP. There are many QoSissues experienced by packet-switchednetworks that do not affect circuit-switchednetworks.The Internet is a best-effort network, wherevariable latencies and dropped packetsoccur. Because voice services require real-time transmission, VoIP often results in aheavily degraded QoS.

There are four types of delay in IPnetworks: propagation delay, networkdelay, accumulation delay and processingdelay. The overall delay is the aggregate ofthese components:

Propagation delay is that which is causedby the signal having to travel a distance,and is therefore a function of distance. It isgoverned by the laws of physics and cannotbe overcome.Network delay is a function of the capacityof the pipes in the network and theprocessing of the packets as they transit thenetwork. The delay associated with jitterbuffers (see below) is considered part ofthe network delay.Accumulation delay depends on the type ofvoice coder used. It is caused because afinite amount of time (varying from asingle sample time to several msecs) isneeded to collect a frame before theprocessing begins.Processing delay is caused by the actualencoding and collection of encoded samplesinto a packet for transmission. This alsoapplies to decoding at the receiving end. Itis a function of the coding algorithm usedand the processing time.

Jitter is the variable inter-packet timingcaused by the network a packet traverses.It can be considered the standard deviationin delay. Not only is it impossible topredict or control how many hops a packetfrom a VoIP call will traverse, packetsfrom the same call are often assigneddifferent routes, with varying numbers ofhops and different traffic volumes along theway. Because of this, packets from thesame source can experience differentdelays on the way to their destination.

Jitter is removed by buffering fast packetsin order that the slowest packets arrive intime to be sequenced correctly. Thiscauses additional delay, as the buffer mustaccount for all packets, thus delaying themall by up to the delay of the slowest. Theconflict of minimising delay and removingjitter has resulted in various methods ofdynamically adapting the buffer size tomatch the time variations. This minimisesthe delay associated with jitter whilepreventing buffer underflow [19].

14

If the one-way delay exceeds about200msecs, the two speakers will adopt amode whereby one speaks, the other listensand pauses to make sure the speaker isfinished, then speaks. Usually these pausesare ill-timed; thus speakers will end upstepping on each other's speech.

Present IP networks are best-effortnetworks and do not guarantee service.When a router becomes overloaded, it mayintentionally drop packets to relieve thecongestion. With traditional data traffic,there are error-checking methods built intothe protocol (Transport Control Protocol,TCP) to address these situations andmaintain data integrity. This protocolrequires some overhead not conducive toreal-time traffic, and has not beenimplemented for voice transport. Instead,UDP (User Datagram Protocol) is used asthe transport level protocol. It requires lesscapacity, but does not guarantee delivery ofthe data to the destination, nor delivery ofpackets in the right order, nor does it avoidduplication of packets. It does, however,guarantee data integrity.

Fortunately, uncompressed voicecommunications tend to be highly tolerantof dropped packets. The human ear cannotdetect the loss of a packet in an audiosequence if an adjacent packet is played inits place. It is only when several packets ina row are dropped that a noticeabledegradation of quality occurs. However,many voice compression mechanisms resultin the transmission of the changes (actualor predicted) between adjacent signals,rather than the whole signals themselves.Hence the loss or corruption of a packetcan result in propagated errors, hugelydegrading the quality of voice or video,often beyond recognition.

The use of UDP has a further challenge inthat many networks protected by firewallsdo not permit UDP traffic. For data topass through many firewalls, some sort ofregistration must occur between sender andreceiver. UDP, having no error correctionor retransmission associated with it, neednot set up any such registration betweensender and receiver. This gives the needfor an open port on the client, which resultsin its blocking by the firewall.

When a two-wire telephone cable connectsto a four-wire exchange interface, a circuitcalled a hybrid is used to convert betweenthe two wires and four wires. Echo iscaused by signal reflections in a hybridcircuit. Although the circuits are veryefficient in their conversion ability, a smallpercentage of the energy is not converted,and is reflected back to the caller. This isknown as echo. If the speaker is close tothe PBX or the telephone switch, the echois sent back so quickly that it is hardlynoticeable. As echo reflection timeincreases, the echo needs to be cancelledout. DSPs listen for the echoed signals andsubtract a prediction of them from thespeaker’s audio signal. This is particularlyimportant with VoIP, since echoed signalstend to have delays long enough to benoticeable, even annoying to the speaker.Although echo-cancellation is used, it isoften insufficient as only a prediction of theactual echo is removed.

15

Quality problems are the most importantbarrier to the development of VoIP. Untila QoS management mechanism isavailable, this will remain a majorrestraint. Unreliability will also be causefor concerns in areas such as public safety.PSTNs are usually well engineered andvery reliable. General services are rarelyunavailable, and emergency services areextremely reliable. The Internet can beslow or unavailable for a number ofreasons, and data can get lost or corrupted,making it extremely unsuitable as adependable network.

Inability to connect, and slow traffictransfer are a result of network bottlenecks.There is no single bottleneck on theInternet; rather, each junction on theInternet can create a performancebottleneck.

The first bottleneck (for dial-upsubscribers) is the connection to thetelephone switch. The average phone calllasts about 3 minutes, while the averageInternet user is online for 20 minutes [20].The PSTN was not designed for calls ofthis length, and the abnormally longholding time is resulting in the absorptionof much capacity of the switch, causingmany call attempts to fail. The phone linealso has significantly inferior capacity tothe rest of the Internet. Because the overallconnection is only as fast as its slowestpipe, the last mile (from the telephoneswitch or service provider to the premises)usually slows down traffic transferdramatically.

The second bottleneck is the connectionbetween the service provider and theInternet backbone. As these connectionsapproach overload, some users may bedenied access, while others will experienceresponse time problems.

The third and fourth bottlenecks are in thebackbone itself. The former is thebackbone bandwidth. Although it hasextremely high capacity, the dramatic

increase in demand has rendered thisinsufficient. The latter of the twobackbone bottlenecks is the routers thatreceive and forward the data packets. Thespeed of routers has grown since thebeginning of the Internet explosion, but thisgrowth has not been able to keep up withthe growth in Internet traffic, and thusdelays result.

Finally, the original Internet backboneinterconnection points are becomingseverely overloaded. Over the last fewyears, the main exchange points of theInternet backbone, the Network AccessPoints (NAPs), have become congested andmany providers have chosen not to usethem anymore. It is also the case that theyhave not received funding from theNational Science Foundation since 1998,and there has been a move to privatelyoperated exchange points. This has meantthat, as many providers enter privateinterconnection agreements, much trafficdata is no longer available. The operatorsand providers involved have no obligationto disclose this information.

All of the above contribute to theunreliability and delay of the Internet.

The existence of legacy networks is arestraint on the growth of IP networks,particularly within enterprises. Compellingbusiness cases are needed to replace voicenetworks that work perfectly and offerbetter quality than IP networks.

A variety of interoperability problems arepresented to VoIP vendors, and the marketwill push hard until these are resolved.Current interoperability exists only at theclient software level. This enables H.323clients to communicate with othertelephones through a H.323 gateway.However, no standard yet exists that allowsgateways from different vendors tointeroperate. This is slowing the adoptionof IP by telcos [21].

Network vendors must supply gatewaysand gatekeepers (the software that provides

16

call control and address translation) thatare interoperable. Service providers willnot, in the long run, deploy large-scaleproprietary solutions. Gatewayinteroperability is not fully addressed bythe H.323 protocol, but is expected to beaccomplished by protocol extensions suchas MGCP (Multimedia Gateway ControlProtocol). MGCP is used for controllingVoIP gateways from external call controlelements called call agents or mediagateway controllers. The call controlintelligence is outside the VoIP gatewaysand handled by the call agents. In theMGCP model, the gateway focuses on theaudio signal translation function, while thecall agent handles the signalling and callprocessing functions. Consequently, thecall agent implements the signalling layersof the H.323 standard, and presents itselfas a H.323 gatekeeper. Several venturesare addressing interoperability at this level,with a Lucent alliance developing agatekeeper that enables interoperabilitybetween gateways of different vendors[22].

The problem of integrating VoIP gatewayswith existing PBXs, switches and routersoften results in a complex configuration.This is due to a lack of experiencedtechnicians and the closed architectures ofmost PSTN network elements.

The VoIP community has not yet agreedupon a common Call Detail Record (CDR).This would make it possible for onenetwork provider to access and settle withanother network provider or owner.Without a common CDR, the areas inwhich ITSPs operate will be limited toregions where they have their own IPgateway.

Standard applications for billing andOAM&P (Operations, Administration,Maintenance and Provisioning) are notavailable, forcing users to rely on onevendor for a complete solution. This is agreat restraint, especially if proprietaryapplications cannot be integrated withexisting operations support systems.

Scalability is also a problem here becausethere are many needs in this area. The portdensity of each gateway can be easilyincreased, as can the overall number ofusers due to the distributed nature of theInternet, however sufficient business,service, network and element managementtools and processes are not available,making scalability of general operationsdifficult [21].

When technical problems are overcome, anacceptable voice quality will be available,probably exceeding that of the PSTN.There will be widespread availability ofVoIP without having to subscribe to abasic Internet package, and truly scaleable,interconnected services will becomeavailable [23]. Much development andmaturing will need to occur before thisbecomes reality.

7. Independent Providers in theVoIP Market

Due to the entry of the telcos, and thenumber of new entrants to the VoIPindustry, it is expected that there will be acertain amount of market squeeze as theindustry matures. The ability to survive inthis market will be, in large part,determined by the competitive strategy of aprovider, and its ability to change rapidlyin the dynamic marketplace.

As in any market, those who grow largeenough to generate economies of scale arein a position to survive, as are those findinga niche market to serve.

To the advantage of independent providers,time to market has been very short forInternet products and services. Innovationin traditional telephony has never been apriority, as most providers were protectedmonopolies, so new services have generallynot been promptly deployed. Data serviceproviders realise how important it is tobring services to market quickly, as acompetitive advantage. Their ability to dothis will be seen as a great strength over

17

telcos in this market. Their dynamism interms of technological upgrade andbusiness processes will similarly beadvantageous.

These companies, although they may beyoung, have been in the business of packetnetworking and have accumulatedexperience in this area. Telcos may find itdifficult to transfer their knowledge topacket switching.

Independent ITSPs are willing to take risksbecause they have little to lose. Facilities-based providers, such as telcos and nextgeneration telcos, have invested enormoussunk costs in their infrastructure, and willtherefore be more conservative in theirchoice of activities. Those that haveinvested little more than some venturecapital can take greater risks, and oftenthese will pay off.

As there will be a huge number of newentrants in this market, there will be greatpotential for joint ventures among thesecompanies, to increase geographic scope,customer base, business knowledge andtechnical know-how. There will also be theopportunity to partner with suppliers,bringing billing experience and highreliability. Traditional suppliers alreadyhave relationships with telcos, but datavendors incorporating voice capability intotheir products will want to partner withcompanies focused on providing VoIPservices.

There is also the opportunity to causedisruption to the traditional telephonyindustry and negatively affect incumbenttelcos. Those telcos who have not reactedyet to VoIP may be leaving themselvesvulnerable in the long term.

In spite of the above, there are manydisadvantages to being a small,independent provider. Although youth hasled to strengths such as dynamism andwillingness to take risks, it also results inan overall lack of experience. Businessexperience needs to be accumulated over

time, and billing systems and sustainablebusiness models need to be developed.Most independent ITSPs will have muchdata experience, but little or no voiceexperience. Although some packetswitching knowledge will transfer fromdata to voice, it cannot be assumed thatthis knowledge will be sufficient whendealing with a whole array of new services.

Dominant carriers tend to already haverelationships with established companies,and prefer to do business with them.Companies like to do business with othercompanies of the same size, and this willdisadvantage independent providers in theareas of interconnection and bundling ofvalue-added services. Smaller providerssimply cannot operate without the co-operation of large carriers, and to a certaindegree will have to accept the market ratherthan shape it.

Regulation can be considered both a threatand an aid to any Internet serviceproviders. Lack thereof has allowed theInternet to flourish, suggesting that anysort of regulatory intervention would stiflegrowth somewhat, putting smallerproviders at a disadvantage. However, asdiscussed above in Section 2, generaltelecoms regulation has provided benefitsfor small providers as well as threats.

Incumbent providers also pose a threat toindependent ITSPs. Most have realised thepotential of IP and have acted accordingly.They are investing in VoIP equipment, andsome have already begun to offer VoIPservices.

To sum up, VoIP is growing rapidly, andwill be huge. There are many possibilitiesfor established companies and for start-ups. However, all such markets offershort-term potential and long-term marketsqueeze, sympathising only with those whohave succeeded in their strategicpositioning and relations, both customerand business.

18

8. Telephone Companies and VoIP

Telcos are starting to take the challenge ofIP telephony seriously. Although it can beseen as a threat, it is also a hugeopportunity. Incumbents have a naturaltendency to work on improving existingtechnology and operations within existingbusiness models, rather than embracingnew technologies and industrial change, notwishing to cannibalise their lucrativebusiness.

Telcos own the local loop infrastructure,and tend to own backbone networks; thusthere are opportunities for them to leveragerevenues from each part of the network,either as capacity providers or as retailservice providers.

Within telecommunications, IP gave newcompanies a chance to enter the market byoffering services that posed no major threatto existing traditional operators. However,as technologies have matured andprogressed, a new range of applicationsbecoming available are rapidly threateningtraditional operators’ core business, themost notable, of course, being VoIP.

VoIP, originally considered too unreliablefor mass-market development, gives newentrants an easy and cost-efficient way tocompete with incumbent operators throughthe current price arbitrage associated withInternet communications. Along with thisprice advantage, VoIP offers a platform forthe integration of voice and data, richerapplications and increased functionality.The opportunities presented by the Internethave resulted in many new entrants –without traditional business investments toprotect – wishing to exploit the commercialpossibilities of VoIP. The short-termeconomic advantages available will allownew entrants to build up their customerbase and know-how (both technical andbusiness), leaving them to gain potentialfor the longer term.

Incumbent telephone operators need tothink very seriously about IP and its related

services. Several have already begunexperimenting with VoIP, tackling theproblem of undercutting their own pricesbefore other ITSPs do. Future-thinkingincumbents have been quick to respond tothis challenge. Deutsche Telekom,Germany’s incumbent telco, has becomeEurope’s largest ISP, and due to the highnumber of new entrants planning to useVoIP, has purchased a stake in VocalTec,and has begun offering a VoIP servicecalled T-NetCall. The company hasexpanded its service outside of Germany,installing gateways in Japan, the UK andthe US, and is taking advantage of fullliberalisation in the European market.

In the US, MCI is the most notable of thetraditional operators who have entered theVoIP race, having already launched atechnology enabling integrated voice anddata transmissions for call centres. Moreservices are being planned for the future asMCI continues to expand on its telephonyrevenues.

Other large operators in the US haveseemed excited about the possibilities ofIP, yet have been reluctant to offer theseservices in their own domain. AT&Toffered commercial phone-to-phoneservices in Asia-Pacific (particularly inJapan) for nearly a year before commercialtrials began in the US. AT&T’s Japaneseventure includes 27 Japanese companies,and offers business and residential VoIPservice throughout the country, with callsoffered at up to 80% cheaper than standardservices.

Because long-distance and internationaltariffs in the Asia-Pacific region aregenerally high, and call volumes arerapidly expanding, many new entrants aretrialling VoIP there. This has resulted in aresponse by the home incumbents offeringsuch services. Japanese companies havebegun to offer prepaid IP phone-to-phonecalling card and FoIP services, whileTelstra in Australia has been underpressure to offer VoIP services betweenSydney and London.

19

8.1 Telcos in the VoIP MarketIt is evident that the incumbent telcos aremoving into the VoIP market, and havesignificant advantages and disadvantages inentering a market neighbouring their own.Their strengths, weaknesses, opportunitiesand threats are identified:

Strengths: telcos own the local loop, andoften own large parts of the Internetbackbone. This is a huge advantagebecause a high percentage of costs can beeliminated. Telecommunications, ingeneral, are becoming very customer-centric, meaning that the most importantpart of the value chain is that nearest to thecustomer. In owning the local loop, telcosmanage to be near the customer, whetherthey are providing services or leasing theirinfrastructure to others.

Incumbent telcos have been providers oftelecommunications services for manydecades, and as a result have accumulatedknowledge of network management,dealing with large customer bases, and theoutlook of a multinational company.General business practice is not abundantin start-up companies, many of whom havelittle more than some technical knowledgeand venture capital.

Telcos already have a large customer base,and due to loyalty and convenience, manycustomers will be happy to accept VoIPservices from this company rather thanlook further afield. They also have namesthat are known. They have huge capitalresources, both existing and potential, andare big enough to undercut many othercompanies on price.

Weaknesses: the weaknesses telcosexperiencing on entry into the VoIP marketmay be due to expectations thatcommunications provision is the sameacross the board. They may fail tounderstand that the new networks andtechnologies with which they are dealingare not wholly analogous with those theyhave become familiar with. This could

lead to the use of business and otherprocesses that are unsuitable or outdated.

The business model associated with telcosis not based on marginal costs, as will befor VoIP provision. Traditional telephonyhas also relied on cross-subsidisation andaccess charges so that universal servicemay be provided. The VoIP model will bebased on marginal costs, which areminimal compared to fixed costs. Theability of large incumbent companies toadapt to this new scheme is arguable, asthey have always had difficulty with rapidchange. However, if telcos do manage tomove away from their traditional costmodel and lower prices towards marginalcosts, the attraction based on pricearbitrage will be minimised.

The idea that they are cannibalisingexisting business may retard some largeoperators. Those that have enjoyedmonopoly status in the past are alreadysuffering at the hands of new entrants, andmay be averse to further reducing short-term margins, even if this may result in along-term downfall.

Because the regulation (or lack thereof) ofVoIP is questionable (see Section 2,Regulation of VoIP) many telcos havetaken the attitude that the regulators willdeal with their threat of ITSPs. This mayresult in telcos being unprepared for thecoming market, and being left to playcatch-up when they realise the extent ofVoIP uptake.

Opportunities: telcos entering the VoIPmarket will experience many of theopportunities of ISPs and vendors, such asmergers and acquisitions of small andmedium-sized companies. Deals andalliances may be formed to generateeconomies of both scale and scope. Manyof these joint ventures and take-overs havebeen evident already, as telcos see this asan easy way of entering the VoIP market,and at the same time can acquire customerbases and technical expertise.

20

Large telcos owning much of their networkwill have the opportunity to leveragerevenues at all points of the network: localloop, leased lines, backbone, as well asavoiding having to pay others to carry theirtraffic. It appears that this will be thegreatest advantage of the telcos in thismarket. As the NAPs break down,interconnection agreements are becomingcommonplace, and those that stand to gainthe most from these are the owners andoperators of the networks.

Because they have much experience in thetelecommunications industry, and becausethey can offer bundled services (one-stopshopping – phone, Internet access, onlinebanking etc.), they may be able to squeezemuch of their competition on price andservice offerings. Smaller providers,although in a prime position to serviceniche markets and to change rapidly, aresimply not in a position to be able to offerbundled services without the co-operationof large operators.

Threats: because telcos will enter the VoIPmarket at a high level, there will be notonly competition amongst themselves, butcompetition from the established players.These players know the market and knowthe customers, unlike telcos who may makethe mistake of thinking that theirknowledge will transfer to this market.The accumulated knowledge of large ISPsand ITSPs will need to be learned bytelcos.

Whilst the opportunity for telcos toexpand, both in terms of size and expertise,exists in the form of acquisition andmergers, similar activity is occurringamongst other ITSPs, making the survivingcompanies more dangerous.

The emergence of next generation telcoshas shown the necessity for such a strategicfocus. Telcos providing bundled servicesmay lack such a dedicated focus, and as aresult may not be able to service theircustomers adequately.

Regulation in this area is highly debatableand thus must be considered a threat.

Fig 1 Telco SWOT Analysis [source 25]

There are three basic strategies that telcoscan adopt in the face of VoIP.

The first involves staying still and hopingthe VoIP phenomenon will never take off.This was the general strategy among telcosuntil recently, expecting that voice qualitywould never be sufficient on such ageographically distributed packet-switchednetwork as the Internet. There was also anexpectation that price differentials wouldeventually dissolve. Although this mayprove true, the price arbitrage currentlyfuels VoIP adoption, but it is theavailability of enhanced functionality thatwill take over as PSTN prices fall.

The quality of voice on the Internet willeventually exceed that of the PSTN, andother current PSTN advantages, such asthe integration of intelligence into thenetwork, are already entering the Internet.Nortel Networks has announced a nextgeneration SS7/IP signalling gatewayapplication, which is to deliver large scale,carrier-class PSTN/IP integration [24].

The above approach would maximiserevenues in the short term, however asprices are forced down by competitionfrom new entrants and VoIP providers,revenues would slip similarly. At the sametime, innovative providers will gather acustomer base and establish credibility.

Strengths Weaknesses

Opportunities Threats

Own infrastructure (backbone and local loop)Global presence and/or outlookHuge capital resources, both existing and potentialLarge existing customer baseName recognitionUnderstanding of scaled network managementAbility to bundle different services (1-stop shopping)Big enough to undercut on price

New networks not wholly analogousAversion to cannibalising existing lucrativebusinessBusiness processes outdatedInability to adapt to volatile market conditionsLack of corporate understanding for the newcompetitive environment

Leveraging revenues at all points of the networkUnified service and billing proposition forcustomerSqueeze competitors on price and service offeringsSign deals and alliances to generate economies ofboth scale and scopeRelationships with large ISPsAcquire small/medium companies

Effective marketing and customer service by competitionConsolidation amongst competitionConsumer choiceEmergence of ITSPsRegulation

21

Operators wanting, at a later stage, to enterthe IP arena, will find themselves in a raceto catch up, and with a tarnished brandimage.

Operators seeking to become early movershave to make the difficult decision to offerlow priced services in direct competitionwith their existing business. This strategycould take advantage of their known namesand available capital to obscurecompetition. Telcos would be in a positionto offer differentiated services over theInternet (or other IP networks) or thePSTN, depending on the requirements andsensitivities of the customer. They couldalso move into new markets as newentrants, as AT&T have done successfullyin Japan, whilst learning about these newmarkets and services.

This approach seeks to maximise revenuesin the long term by compromising someshort-term margins. However,segmentation of the market and the offeringof differentiated services does notnecessarily mean revenues will dropsignificantly in the short term.

The final strategy involves operatorslooking to climb the value chain byexploring value-added VoIP services andoffering them to their corporate customers.This could involve movement into any ofthe primary areas of value-added servicesgenerally associated with ISPs - content,systems integration, web hosting. Becausethese services are very far from thecompanies’ traditional strategic focus andcore competencies, many have failed to besuccessful in this area.9. VoIP Market Segmentation

The VoIP market, in its immature stage, ishighly segmented, with particularfunctionalities and enhanced servicesespecially suited to certain customersegments. An industry that began byoffering simple low-quality services toInternet-savvy hobbyists is now buildingcustomer bases in various areas oftraditional telecommunications and data

communications, and will soon be able tooffer mass-market and ubiquitous servicesto each member of the public.

The emphasis so far has been in two mainareas – business use and long-distance/international use. Current VoIPmarkets are heavily biased towardsbusiness usage, but consumer markets aredeveloping quickly as a result of nextgeneration carriers.

9.1.1 Business UsageQuality problems such as latency can beminimised over a local network such as acorporate LAN. For this reason, manybusinesses running IP LANs have takenadvantage of the efficiency of packetswitching. Technical problems are lessdaunting due to the closed nature of thenetwork, which results in greater reliabilitythan on the Internet. This means that delayin processing and reception can beminimised, and calls are closer to PSTNquality. VoIP may also be adopted forinter-office communications (on a WAN orLAN-to-LAN over the Internet), wherequality is less of an issue than whendealing with customers or partners.

Businesses can also set up web-enabledcall centres, so that customers wishing tocontact them by phone can initiate a callwithout having a second line ordisconnecting their Internet session. Theycan ‘click to talk’. If the customer too hasVoIP capability the call can be made overthe Internet without the customer having todisconnect his connection. Otherwise, thecustomer can enter his information, send itover the Internet, disconnect and wait for acall centre agent to call him back.

Shared applications are also possibleacross an Internet connection, allowingboth the caller and the agent to shareinformation and to interact. This couldsignal a radical change in the waycustomers interact with a company, goingfrom a purely auditory experience to onethat can involve speech, informationsharing through web interaction,

22

whiteboarding [glos] and file exchange.Push technology can be used to guide acustomer to other pages on the Internet.Instead of the customer requesting thesepages using HTTP, the agent can ‘push’the information onto the customer’scomputer.

There are also advantages to VoIP callcentres in terms of time. Forms can befilled in directly by the customer, and whenhe connects to the call centre theinformation is sent directly. Also, cookies[glos] can be left on the customer’s PC thatwill identify them to the call centre the nexttime they call, saving both the caller andthe agent time.

Businesses with high incoming toll-free callvolumes can eliminate their toll-freenumber access charges by lettingcustomers call a local number, from whichthey are transported to an internal,centralised network. This appears to be alocal call to the customer, unaware that hiscall has been transported over a packet-switched network, and costs that for thebusiness [26]. Businesses that have hightelephony usage, such as travel agents andfinancial services, will be the target ofmany new entrant VoIP providers.

SOHO (Small Office/Home Office)premises tend to have higher penetrationrates than consumers for multiple telephonelines, PCs, and online services. Their highconnectivity and multiple needs are anopportunity for VoIP service providers.Being already connected to the Internet, avirtual second line can be provided byreception of VoIP and through the PC.Significantly enhanced services, which maybe of great use to small businesses, can beoffered. Small businesses are likely to beless concerned about the lack of QoS onthe Internet, and will be willing to trade offconvenience and quality for lower prices.

9.1.2 Residential UsageInternet hobbyists, the original users ofVoIP still use PC-to-PC calling. However,the primary focus in the residential market

has become phone-to-phone long-distanceand international calling.

International and long-distance residentialcallers are willing to accept lower-qualityinternational calls at a lower price. Theseusers are less sensitive to quality, and donot necessarily require enhanced services.It is the price arbitrage alone that hasattracted them to VoIP. Consumersgenerally tend to be cost-conscious andmore flexible than business customersregarding quality and convenience.

Next generation telcos are moving towardscatering for large cities having highdensities of people with strong foreignlinks. Offering VoIP between, say, SanFrancisco and China, would allow serviceproviders to build up large amounts oftraffic between these locations. Thisbackbone cost reduction seems to be one ofthe main objectives of the next generationtelcos. Gateways can be appropriatelypositioned, and a dedicated line can be setup between these destinations. This mayallow for VoIP without any use of theInternet, and hence overcome significantQoS problems, as capacity can be easilymanaged and long distances can betraversed in just one hop.

9.2 VoIP UptakeFrom country to country, the developmentof VoIP will depend on a number ofcomplex and dynamic issues. Certainenvironments are more inclined towards thetake-up of VoIP for many reasons. Forinstance, in the US, tariffs are lower thanelsewhere, and quality tends to be higher.Therefore, most customers will be reluctantto move to lower quality services when theprice arbitrage is not so evident. On theother hand, in countries with poorlydeveloped infrastructures the cost/qualitytrade-off of VoIP will be considerablymore attractive. The regulatoryenvironment may also have an effect here,as a competitive telecoms environmentgenerally results in lower tariffs. However,such an environment can foster innovation,

23

leading to a quicker uptake of newservices.

The Internet infrastructure and penetrationof Internet usage will affect adoption.Customers already online will be morewilling to try out VoIP, and a well-developed infrastructure will be able togive a higher quality of service. Thepenetration of high-speed local looptechnologies will also facilitate VoIP. Thelast mile is usually the slowest part of anInternet connection, and with theintroduction of xDSL and other high-speedtechnologies, enhanced VoIP servicequality could increase dramatically.

The maturity and sophistication of the ISPmarket in a particular country will be anindicator of the time-to-market for newservices.

Specific routes (e.g. east-coast USA toIreland, UK to south Asia) with highvolumes of consumer traffic will takeadvantage of the low costs of internationalcalling.

10. Market Forecasts

VoIP revenues are currently higher in theUS than in Europe, but this is set to changedue to both the current high tariffs inEurope, and the recently liberalisedmarkets bringing competition. Usage willbe greater in the US, primarily due to thesuperior Internet infrastructure, butmargins will be higher in Europe, allowingproviders to reap greater revenues.

Within Europe, the VoIP market isexpected to develop fastest within maturecommunications markets such as the UKand Scandinavia, and in countries withhigh PSTN tariffs, such as Italy and Spain.

Fig. 2 VoIP Service Markets 1997-2002, source [5]

According to Frost & Sullivan, the totalVoIP market will experience compound

growth of 149% for the next few years,while minutes of use will increase byapproximately 220% annually.IDC (International Data Corp.) hassuggested the market may be worth asmuch as $24.4bn by 2002, withinternational calling generating more than83% of this revenue, whereas Killen &Associates estimate revenues will reach$63 billion by 2002.

Consultancy Phillips Tarifica estimatesthat AT&T will lose $620m to $950m ininternational calls by the end of 2001 dueto VoIP. Digital estimates that VoIP couldcost traditional phone companies $8 billionin lost revenues over the next four yearsand may be able to eliminate the profits ofU.S. long-distance carriers by stealing just6 percent of U.S. telephone traffic [27].According to Forrester Research, by 2004,ITSPs will offset the incumbent carrierrevenues by $3 billion in domestic longdistance, $2 billion in spending and $1billion in savings by end users. ForresterResearch also predicts that prepaid callingcard traffic will drive 50% of VoIP enduser spending by 2002, reaching $1.4billion by 2004 [28].

Probe Research predicts that IP networkswill carry 7% to 11% of the world'sinternational voice traffic by 2002, 34% ofthe U.S. domestic long distance traffic, and10% of the world's fax communication.They also suggest that as much as 50% ofVoIP minutes will be additive, i.e. minutesthe PSTN is not losing [29].

0

500

1000

1500

2000

2500

3000

3500

1997 1998 1999 2000 2001 2002

US

$m US

Europe

VoIP Minutes of Use

0

2

4

6

8

10

12

14

16

1997 1998 1999 2000 2001 2002

Bill

ions

of M

inut

es

US Long-Distance

Europe Long-Distance

Us International

Europe International

24

Fig. 3 VoIP Minutes of Use 1997-2002, source [5]

11. Conclusion

Although competition and demand willdetermine the market, a number of factorsare clear in order that VoIP gainwidespread acceptance.

QoS will be a pre-requisite for mass-market VoIP services, particularly businessservices. It is critical that networks havethe ability to prioritise traffic to minimisenetwork delay, and different service levelsmust be available for different classes ofuse.

The market is used to the PSTN, and isreasonably satisfied with it. Users requireseamless integration of the PSTN and IPnetworks. They require services that arealways available and reliable, with the highquality and enhanced calling features theyhave come to expect. The average userdoes not care about the networktechnologies involved as long as hisexpectations are met and modifications tohis behaviour need not be dramatic.End user equipment needs to become moreuser-friendly; reliability must approachthat of the PSTN; interoperability amongstvendors needs to be accomplished throughan open architecture and industrystandards; and accessibility needs to beextended beyond the telephone to the faxmachine, PC, and other compatibledevices.

Once the price advantage dissolves, themarket will be service-driven withenterprises pushing these services. Tariffsare lower than those for the PSTN due to atoll bypass, but it can be argued that thecosts required to build an IP network withthe reliability and features of the PSTNmay yield only marginal cost benefitsrelative to the PSTN. Service providers,through business-oriented services andapplications, can achieve higher margins.Those building dedicated IP networks are

in a prime position to grab a share of thebusiness market, as they can offer QoSguarantees over their managed privatenetwork.

In Internet industries, the early moverusually wins vital market space. Netscape,for instance, a start-up in 1994, almostimmediately became the largest provider ofweb browsing software, and did so becauseit saw the opportunity before Microsoftand other software houses. Equipmentvendors need to enable service providers tooffer business telephony features thattransfer the features of the PSTN to the IPworld, benefiting both themselves and thelatter.

The long-term view (5-10 years into thefuture) sees quality issues as a thing of thepast. By that time there will be no need tomake the economic case for VoIP, asPSTN tariffs will have dropped in responseto competition. Users will opt for VoIPbecause it is more efficient, has improvedfunctionality, and is an altogether bettersolution to serve a technology-savvymarket’s needs.

25

Glossary

ATM – Asynchronous Transfer Mode, amultiplexing and switching technology used totransport small fixed-length packets, calledcells, over a high-speed network.

Concentrator-Based Gateway – aconcentrator (or aggregator) is a device thatacts as an efficient forwarder of datatransmission signals. A concentrator-basedgateway is one that resides at such anaggregation point.

Cookies – special text files that a Web siteputs on a PC so that it can remembersomething about the user at a later time.

DSP – Digital Signal Processor, a circuit usedto improve the accuracy and reliability ofdigital communications. A DSP circuit is ableto differentiate between human-made signals,which are orderly, and noise, which isinherently chaotic.

Facilities-Based Providers – those that buildand operate their own networks.

H.323 – the standard for the transmission ofreal-time audio, video and datacommunications over packet-based networks.

Independent/Non Facilities-Based Providers– those that do not build their owntelecommunications network infrastructure,leasing capacity from others, or paying peruse.

Local Loop – the wired connection from atelephone company's central office to itscustomers' telephones at homes andbusinesses.

PBX – Private Branch eXchange, a telephonesystem within an enterprise that switches callsbetween enterprise users on local lines whileallowing all users to share a certain number ofexternal phone lines.

QoS – Quality of Service, the idea thattransmission rates, error rates, and othercharacteristics can be measured, improved,and, to some extent, guaranteed in advance.

RBOC – Regional Bell Operating Company,one of the US regional incumbent telephonecompanies.

Router-Based Gateway – a node thattranslates between IP and PSTN information,and which resides at a router.

SDH – Synchronous Digital Hierarchy, theinternational equivalent of SONET.

SMS – Short Message Service, a service forsending messages of up to 160 characters tomobile phones that use Global System forMobile (GSM) communication.

SONET – Synchronous Optical NETwork, theUS standard for synchronous datatransmission over optical media.

SS7 – Signalling System No. 7, the globalstandard for PSTN telecommunications.

Whiteboarding - the sharing of a documentor image between by two or more users,allowing them to make notes or changes usingthe drawing capabilities of the particularprogram they are using.

26

Appendix 1: Comparison of Capacities of IP over ATM and IP over SONET

Based on a 155.520 Mbps capacity link running IP packets of size 576 bytes.

Protocol Layer Available Bandwidth(Mbps)

% of Line Capacity % Overhead Addedby Each Layer

SONET 155.520 100 3.7ATM 149.460 96.6 9.43AAL 135.362 87.5 6.41LLC/SNAP 126.937 80.7 1.37IP 125.918 79.6 0

Protocol Layer Available Bandwidth(Mbps)

% of Line Capacity % Overhead Addedby Each Layer

SONET 155.520 100 3.7PPP 149.460 96.6 1.54IP 146.15 95.4 0

source [30]

Based on the above figures, IP over ATM gives 79.6% of line capacity, compared to 95.4% forIP directly over SONET/SDH. The reason for this significant differentiation is that ATM andits related layers provide QoS guarantees.

27

References:

[1] European Commission Directive90/388/EEC on Competition in theMarkets for TelecommunicationsServices

[2] Internet Telephony - The RegulatoryIssueshttp://204.191.126.89/home/mccarthy/web/docs/mt-inter.html

[3] Nokia/Vienna Systems Homepagehttp://www.viennasys.com/

[4] ‘Siemens Introduces IP Telephone’,Zona Research, 15 April 1999

[5] ‘IP Telephony Markets in Europeand the US’, Woll, B., Datamonitor,1998

[6] The IP Pulse Mailing List, GecKoResearch & Publishing, 29 June 1999

[7] ‘Combining Mobile and InternetServices’, Griffiths, N., Dataquest,June 1998

[8] IP Promises a Facelift for Faxhttp://www.soundingboardmag.com/articles/851justf.html

[9] ‘Core Internet Subscription Service.Module 3: IP Telephony, Fax andMessaging’, Mine, H. and Bergeron,J., Probe Research Inc., 1998

[10] ‘IP Telephony Markets in Europeand the US’, Woll, B., Datamonitor,1998

[11] Ruling The Wires - ITU Vs WTOContest Over Telephone Revenueshttp://www.oneworld.org/panos/news/34nov98.htm

[12] The Market for Internet Telephonyhttp://www.firstconf.com/ISP/transcripts/telephony/htm

[13] ‘Core Internet Subscription Service.Analyzing the Industry that Deliversthe Public IP Network’, Vol. 1, No.11, Mine, H. and Zeppetella, M.,Probe Research Inc., July 1998

[14] ‘Korea Telecom Uses SS7Technology to Offer IncreasedServices and Customer Choice for IPTelephony Calls’,Telecommunications TeleViews,Issue 38, Dataquest Inc., September1998

[15] VOIP That Even a Telco Can Love?http://www.bcr.com/bcrmag/06/98p16.htm

[16] Nortel Makes Splash In VOIPMarketplacehttp://www.interactiveage.com/news/news0701-2.htm

[17] ‘Prices May Halve as Europe'sBandwidth Booms’, Cukier, K.,CommunicationsWeek International,19 October 1998

[18] ‘The Factors Affecting the Growth ofVoIP’, McPhillips, E., Hewlett-Packard Technical Report HPL-IRI-1999-003, May 1999

[19] Telogy Voice Over Packet Tutorial,http://www.webproforum.com/telogy/full.html

[20] 'ISPs: From Protocols to the Market',McPhillips, E., University ofStrathclyde, September 1998

[21] ‘World Markets for IP TelephonyEquipment and Services’, Frost andSullivan, 1998

[22] Ascend, Cisco, Clarent, Dialogic,Natural MicroSystems and SiemensJoin ITXC, Lucent and VocalTec inSupport of iNow! InteroperabilityInitiativehttp://www.lucent.com/press/1298/981217.nsb.html

[23] ‘Internet Telephony Gold Rush’,Mines, C., Vernier, B., andVanBoskirk, S., People andTechnology Strategies, ForresterResearch Inc., Vol. 4, No. 12, April1998

[24] Nortel Networks Announces NextGeneration SS7/IP SignalingGateway Application

28

http://www.nortelnetworks.com/corporate/news/newsreleases/1999a/2_2_9999083_IPGateway.html

[25] ‘The Structure and Trends of the ISPMarket’, McPhillips, E., Hewlett-Packard Technical Report HPL-IRI-1999-002, April 1999

[26] Voice Over IP: How it Works andWhy You Should Carehttp://home.zdnet.com/products/content/articles/199805/telephony.works/2.html

[27] Internet Telephonyhttp://www.asia-pacific.digital.com/info/LIW06W/

[28] ‘Internet Telephony Gold Rush’,Mines, C., Vernier, B., andVanBoskirk, S., People andTechnology Strategies, ForresterResearch Inc., Vol. 4, No. 12, April1998

[29] ‘Voice over the Internet, The NextTelecom Battleground’, Mine, H.,Probe Research, October 1997

[30] ‘Comparison of IP-over-SONET andIP over ATM Technologies’, TrilliumDigital Systems, Inc., November1997

[31] WhatIs?http://whatis.com/

[32] The IP PBX Comes of Agehttp://www.phonezone.inter.net/tutorial/ip-pbx.htm

[33] ‘IP Telephony: Opportunity orThreat?’, Lakelin, P.,Telecommunications, Volume 32,Number 10, November 1998

[34] ‘Strategic Positioning and MarketOpportunities forTelecommunications Operators in theInternet Market’, Bouchoris, P.,British TelcommunicationsEngineering, Volume 17, Number 2,1998

[35] The Impact of IP as a TransportProtocol for the Traditional PSTNNetwork Providers

http://www.firstconf.com/ISP/transcripts/overip.htm

[36] Internet Telephony: A Threat toTelcos?http://www.tmcnet.com/articles/itmag/FirstQ/q1actioninfo.htm

[37] Packet Telephony: Long DistanceService for ISPshttp://www.dagaztech.com/warp/public/779/servpro/solutions/voip/ptels_bc.htm

[38] ‘Core Internet Subscription Service.Analyzing the Industry that Deliverthe Public IP Network’, Vol. 1, No.12, July 1998, Mine, H., Zeppetella,M., Alsman, J. and Dela Cruz, M,Probe Research, July 1998