Voice over IP and the Next Generation Network
Response to the ART consultation on Internet Telephony

April 14, 1999

This document contains Level 3 Communications’ response to the ART public consultation document on " Internet Telephony ". Comments or questions can be sent to :

Jo Van Gorp, Vice-President Legal & Regulatory Affairs
Bruno Vanneuville, Manager Regulatory Affairs

Level 3 International
Marsveldplein 5 (Bastion Tower)
1050 Brussel
Belgium

Fax : +32 2 400 52 00
Tel : +32 2 400 50 00

Email : jo.vangorp@Level3.com
bruno.vanneuville@Level3.com

CONTENTS

Executive summary

Chapter one – about Level 3 Communications

Chapter two – Voice over IP and the Next Generation Network

Section 1
Convergence and the concept of Next Generation Networks
Section 2
Towards carrier-grade VoIP

Chapter three – regulatory observations

Legal issues concerning Internet Telephony
Licenses
International context
Numbering
Interconnection and interfaces
Other regulatory issues

Conclusion


Executive summary

Level 3 Communications S.A. (" Level 3 ") would like to congratulate ART on the publication of its consultation document on " Internet Telephony ". Together with the European Commission’s notice on the " Status of voice communications on Internet under community law ", this document represents the first major effort in Europe to assess the regulatory implications of what is to be considered as a fundamental shift in telecommunications.

Level 3 agrees with the ART that " Voice over Internet Protocol " (hereinafter referred to as " VoIP ") is " part of the broad-based evolution in communication technology ", which " may lead to the convergence of voice and data services, and bring into question the separation of the legal frameworks governing them ".

Level 3 further understands, and applauds, that ART’s wish is not more than " to begin gathering information to enable an in-depth study  of the matter ". From the outset, it should indeed be clear that industry thinking about VoIP and, more in general, network convergence is still evolving at a tremendous pace and that, as a result, integrating these evolutions " smoothly " into the (French) regulatory framework cannot take place overnight.

Nonetheless, while this suggests to insert a waiting period before embarking on a full-blown restructuring of telecommunications law, Level 3 would like to underline that, precisely because of this tremendous pace of evolution, VoIP is evolving from a hobbyist nicety into a carrier-grade basic telecommunications service today, as the " Next Generation Networks " supporting this, and other, advanced servics, are being built now. Therefore, regulators have to make sure that existing legislation poses no obstacle whatsoever in this respect. In our view, and as the ART itself appears to suggest, smooth integration of VoIP and network convergence into the (French) regulatory framework requires the immediate and definitive removal of any remaining technology-dependent references in the law or in the interpretation and implementation thereof.

In the following chapters, we will present a few elements to substantiate this viewpoint and investigate the regulatory consequences. However, what applies to the industry as a whole also applies to Level 3 : our understanding of technology, regulation and the market keeps evolving and we do not attempt to give final answers here.

Chapter one provides background information on Level 3.

Chapter two sets out Level 3’s view on the broad evolution of the telecommunications industry towards convergence, against which to judge the concept of VoIP. This chapter also covers sections I - III of the consultation document.

Chapter three deals with regulatory issues.

Chapter one – about Level 3 Communications

Level 3 Communications, Inc. is a communications and information services company that is building an international advanced IP-based network. On April 1, 1998, Level 3 common stock started trading on the Nasdaq National Market under the symbol LVLT.

Level 3 Communications, Inc., through Level 3 International, Inc., has full subsidiaries in Belgium, France, Germany, the Netherlands, the UK and some other countries. Level 3 Communications S.A., which is the French subsidiary, has been granted L.33-1 and L.34-1 licenses and will start offering services in the third quarter of this year.

First International IP-Based Network

The Level 3 network will be the first international communications network to use Internet Protocol (IP) technology end-to-end. Level 3 will focus primarily on the business market, using its IP-based network to provide a full range of communications services -- including local, long distance, and data transmission as well as other enhanced services. Additionally, the company will offer a range of Internet access services at varying capacity levels, and at specified levels of quality of service and security to meet the needs of its business customers.

The company has obtained PTO licenses in most parts of the US and the EU, and is presently offering services in 17 major US cities as well as in London and Frankfurt. By the end of the year Level 3 expects to be operational in 25 US cities and in three other European cities, namely Amsterdam, Brussels and Paris. Initially, services are offered on leased lines, but will gradually migrate to Level 3’s fiber platform as this becomes available. This fiber platform will by then encompass 50 local networks and a 25 000 km long distance network in the US, another 21 local networks in Europe and Asia and a 3 000 km long distance network in Europe. Level 3 has also secured IRU capacity on the Trans-Atlantic and Trans-Pacific routes.

Fundamental Shift In Technology

We have shaped our strategy to build an IP-based network from the ground up because of what we believe is a fundamental shift that is occurring in the communications industry - a shift as important as that from the telegraph to telephone or from the mainframes to the personal computer. It is a shift that we and a growing number of industry experts believe will change the way businesses communicate. That change is a move from the traditional "circuit switched" networks that were designed primarily for voice communications - and which have served customers for nearly a century - to newer "packet switched" networks using IP, or Internet Protocol. This new technology makes it possible to move information at a much lower cost, because packet switching makes more efficient use of the network capacity.

Major telecommunications carriers are aggressively moving to add IP-based switching technology to portions of their networks. But because of the immense investment these companies have in their existing legacy networks, they continue to retain extensive networks based on the older and less efficient circuit switching technology.

We, on the other hand, believe that we are well positioned for the fundamental shift to the new technology because we do not have an investment in, or commitment to, the older technology. We will therefore build our new network from the ground up with the new technology. Equally important, we plan to design our network to be fully upgradeable, so as the technology evolves, so does our network.

The advanced IP-based network will enable business customers to benefit from the lower cost and service offerings made possible by Internet Protocol technology.

Chapter two – Voice over IP and the Next Generation Network

Level 3 agrees with ART’s fundamental observation that the emergence of VoIP may prove to be a catalyst in restructuring the telecommunications market and that it is part of a broader industry evolution to convergence of voice and data services. Therefore, we will first review this evolution to convergence, and the associated concept of " Next Generation Networks " before going into the details of carrier-grade VoIP, especially as an accurate understanding of the former will influence the regulatory perception of the latter.

Section 1

Convergence and the concept of Next Generation Networks

There has been, and, probably, continues to be, a tendency to regard IP based services as " enhanced ", as opposed to " basic " services such as those provided over the reliable, good old circuit switched telephone network (PSTN). However, the emergence of " Next Generation Networks " (NGN) should start to change this perception. Most analysts now agree that in the next decade or so the NGN will replace the PSTN as the ubiquitous platform over which all basic public communication services are provided.

Why is this happening now ?

For most part of the century, voice traffic has been predominant, but today it represents an ever diminishing percentage of communications traffic as we witness much sharper growth in data and fax transmission. It is expected that data will account for more than 95% of the traffic on the public network by the year 2005, and according to some analysts, 1998 will be remembered as " the year in which data exceeded voice traffic for the first time ". We are not so much interested in the statistical discussion of this trend, which would lead to far and add no real value, but rather in the undeniable impact of this trend on the evolution of network architecture.

Broadly spoken, two strategies were conceivable for dealing with this explosion of data traffic : one, for evident reasons favored by incumbent operators, was to carry data over the existing ubiquitous circuit switched and voice-oriented network ; the other, pioneered by new entrants and promptly followed by incumbents, was to start deploying data-friendly packet switched networks. Today it is undisputed that the latter approach will ultimately prevail for at least the following reasons .

Packet switching uses resources more efficiently

The inherent advantage of packet switching over circuit switching is well-documented. Circuit switching technology dedicates a fixed amount of capacity for the entire communication. So, a telephone call ties up an entire circuit – or portion of the network – for the duration of the call. Data networks, on the contrary, put information into small packets that are sent through the network without predetermining the path the packet will take and " without utilizing physical resources in excess of what the packet requires. This sharing of resources provides much higher return on invested capital. "

Silicon economics 

On top of that comes a " virtuous cycle " of elastic demand and decreasing prices. IP-based networks are already less expensive today for non-timing sensitive information. For example, Level 3 has calculated that it would cost approximately US$ 27 for a provider with both local and long distance facilities to move a 650-megabit CD ROM worth of information from New York to London (the equivalent of eight encyclopedia volumes).

To move the same amount of information over an IP network would only cost the provider about US$ 2.00. This is the fundamental cost difference or advantage that IP enjoys today. That difference should grow over time because IP technology is advancing so much more rapidly than traditional telephone technology. According to Peter Sevcik, an analyst with Northeast Consulting Services, circuit switching technologies are doubling their performance/cost ratio every 80 months, while packet switching technologies double this ratio every 20 months. That huge improvement difference " guarantees the triumph of packet switching ".

Convergence 

The recent maturity of VoIP has pavedthe way for the integration of voice and data on a single network, which promises economic advantages to end users, not only in terms of direct transmission costs, but also in reducing the network management costs of running separate and technologically different networks.

Deployment of the " Next Generation Network "

From the above it is clear that we have reached the point of no return at which it is more efficient to run voice on top of packet switched data networks as " just another traffic type ", with IP being the most plausible candidate to substitute the PSTN, given the widespread deployment of IP-based equipment in enterprise networks.

This does not mean that the existing Internet networks incorporate all necessary characteristics that would justify their perception as the basic public networks of today. As Grant Lenahan at Bellcore pointed out, the existing Internet fails to address at least two major challenges :

However, as will become clearer in section 2, companies like Level 3 have started deploying networks and technologies that are specifically designed to meet the demand for QoS and voice parity with the PSTN. In addition, protocols are being developed that ensure seamless interoperability between these NGNs and the PSTN. This is critical to maximise the positive externality effect of interconnection in the presumably extended timeframe during which the NGN and the PSTN will co-exist.

Section 2

Towards carrier-grade VoIP

Today voice parity with the PSTN is achieved over IP-based NGNs like the Level 3 network. This is apparent from a discussion of service types, QoS, underlying network and interoperability, which are the issues raised by ART in the consultation document.

Service types : architecture, equipment, standardisation

Level 3 agrees with ART that most VoIP services take the form of " computer to computer ", " computer to phone " or " phone to computer ", or " phone to phone ". However, we do not think that this practical observation should be taken as a basis for classification of the VoIP universe of services. The specific type of end user equipment used in a VoIP service merely indicates where the PCM/IP conversion takes place but not necessarily contains information on such things as supported type of calls (outgoing, incoming) and degree of interconnectivity.

This being said, it is clear that, on the terminal equipment level, computer telephony integration has an enormous potential for the development of new advanced features. But as carrier-grade VoIP service develops, it may be more appropriate to distinguish between services which provide an alternative to POTS and those services which merely supplement POTS in a limited fashion. The former would require the ability to support both incoming and outgoing calls from and to other subscribers, including those subscribers that are still connected to the PSTN. It is clear that the capability to support this " public voice " type of service depends, not on the type of end user equipment, but on the type of IP voice network architecture.

As far as architecture is concerned, Level 3 agrees with ART that the basic components of any VoIP service are :

Voice gateways are available today from a variety of vendors and are based on the ITU-T H.323 protocol suite. The first version of H.323 was adopted in 1996 as a family of standards that defined the components, procedures and protocols necessary to provide audiovisual communication over LANs and other networks where QoS could not be guaranteed.

H.323 also aimed to standardise the call agent component, but it was generally felt that it failed in this respect because of too much complexity, overhead, lack of scalability and call-set up delay. A number of industry groups formed to develop a new standard for the call agent component, notably the TAC (Telecommunications Advisory Council) which was chaired by Level 3 and proposed the IPDC protocol, and BellCore and Cisco who came up with SGCP as their improvement of H.323. The chances of one uniform industry standard grew when BellCore, Cisco and Level 3, in November 1998, decided to merge their respective standards into MGCP (Media Gateway Control Protocol) and subsequently proposed it to IETF and ETSI Tiphon. It is important to note that MGCP is designed to function as the signalling protocol within the H.323 and, in other words, does not affect the installed base of H.323 conformant gateway and end user equipment.

Quality of service

Quality of service is at the core of voice telephony and, as such, the focal point of the VoIP debate. Perceived poor voice quality over the existing public Internet explains why VoIP often fails to be considered as a basic, carrier-grade service. In reality, the issue is very poorly understood. Poor quality is not an intrinsic characteristic of VoIP and can be managed fairly easily. In this paragraph, we will only deal with the concept and measurement of QoS. We will have a look at how it can be achieved in the underlying IP network in the following paragraph.

Overview

Voice quality is defined by a number of parameters, amongst which the most important are:

Level 3 has performed measurements in order to contrast the customer perceived quality of the Level 3 packet voice network with the standard quality experienced by users of the PSTN. The results are described in what follows.

Call Setup Delay

Normal long distance call setup delay for a 4800 km call on a PSTN are in the range of 1 to 2 seconds. Enhanced services calls such as toll free may take a little longer. A standard 4800 km call on the Level 3 network will be set up within the same 1 to 2 second time frame, so most users of the Level 3 network will not notice any difference.

End to End Latency

In spoken communications, one-way delays must not exceed 250 ms or the delays become bothersome to the parties in the conversation. If delays exceed 250 ms, then because the listening party is hearing what the speaking party said a Ľ second earlier, the listening party may think there is a gap in the conversation at which the listener may respond, only to find out that the speaking party had jumped into the gap and started speaking again. These interruptions make high-latency conversations frustrating.

In fact, humans can begin to notice the delays at about 125 ms of one-way latency. However, even though the delay is noticeable, conversational flow is not severely impacted until latency reaches 250 ms in one-way delay.

On a standard 4800 km long distance call on the PSTN, end to end latency is about 40 ms. That is, it takes about 45 ms for the listener to hear what the speaker said. A worst case domestic US long distance PSTN call covers about 12800 km from Seattle to Miami, and will take about 100 ms in one way latency. For an international call of 19200 km, PSTN delay can be expected to reach 180 ms when most of the distance is covered by fiber optic cables.

Cellullar phone latency, even in the local telephone calls, can reach 125 ms.

The following table compares PSTN and expected Level 3 network latency:
  L3 vs. Circuit
 
One-way latency
 Percent ITU Mean
Call Scenario Existing PSTN Level 3 Network Difference Opinion Score
4800 route mile LD call 45 ms 110 ms 59% worse 3.5 vs. 3.7
12800 route mile LD call 100 ms 175 ms 44% worse 3.2 vs. 3.4
19200 route mile LD call 180 ms 255 ms 30% worse 2.8 vs. 3.0


Notice that as the distance of the call increases the difference in latency between the existing PSTN and the Level 3 network begins to decrease when expressed as a percentage. This is because as the distance increases, both circuit switched and packet networks experience the same transmission network delays, with only minor differences in the latency contributions of the electronics at each hub or switching office.

In terms of user perception of delays, we expect only callers making calls in the range of 11200 to 17600 km to perceive any difference in latency. Even though calls of this distance will provide a noticeable difference in latency, the latency provided by the Level 3 network will not impede the flow of conversation, and will be acceptable to most callers.

Tests were conducted in 1991 by NTT in Japan, and reported in the ITU document G.114, One Way Transmission Time. The tests measured detectability of delay by three different populations: Untrained laymen (businessmen), trained crews, and untrained laboratory employees. Six different conversational styles were measured:

– Task 1: Read out random numbers as quickly as possible in turn.

– Task 2: Verify random numbers as quickly as possible in turn.

– Task 3: Complete words with lost letters as quickly as possible by exchanging information.

– Task 4: Verify city names as quickly as possible in turn.

The table below shows the amount of delay at which the delay was detectable by the test participants. As you can see, for untrained participants delays were noticeable only when one-way transmission time exceeded 300 ms, and then only for complex time-sensitive tasks. Only trained personnel could detect delays in the range of 100 to 200 ms.

Speaker Recognition

Many voice over IP solutions make use of advanced compression algorithms to reduce the bandwidth requirements of a conversation. Use of these compression/decompression algorithms (called codecs) can negatively impact the quality of a real-time conversation by either increasing the latency of the call, or by causing a loss of fidelity in the reproduced sound wave, resulting in odd-sounding audio. For example, using a compression algorithm on audio can make it sound thin or tinny, can make it sound like the speaker is underwater, or can introduce background noise.

On the Level 3 voice network, for phone to phone calls we plan to use only codecs that reproduce sounds with less than 2 ms of algorithmic delay. These codecs include G.711 (the same codec used in the PSTN network), ADPCM, and G.728.

These same three codecs also have excellent records for reproducing quality audio. All three reproduce audio at near toll quality, with ratings of at least 3.8 on the ITU Mean Opinion Score (MOS). Our users will notice no difference in audio quality when comparing our network fidelity to that of the PSTN.

Signal Loss

In the first release of the Level 3 voice network, no signal loss will be experienced by our customers, as our network will not accept any analog line inputs. Signal loss (also called attenuation) is a phenomenon that affects analog signals but does not affect digital signals. Once we do accept analog line inputs we will employ signal processing technology in our Access Gateways to ensure only small levels of signal loss.

Packet Loss

Packet loss occurs when our network fails to deliver data to a network destination. As a next-generation communications carrier, Level 3 uses a data network to transport everything, including phone conversations. Packets are the little envelopes which carry data from one end of the network to the other.

On the Level 3 voice network, the audio received from your phone is broken into 100 little pieces per second, and each piece is delivered in its own envelope (packet) across the network. On the other side of the etwork, the little pieces of audio are re-assembled and played back to the other phone.

We expect to lose less than 1 in 10,000 packets on our Voice virtual private network, causing an imperceptible difference in the quality of voice reproduction for our users. Since each lost packet represents only a 10 ms audio sample, the rare loss of a single packet at a time will not be noticed by customers. In most cases, a single dropped packet cannot even be heard.

Reliability and Recovery

There are no single points of failure on our voice network. All equipment is provisioned redundantly to allow a quick and full recovery in case of component failure. However, failure of some components can cause the failure of all calls in progress through that component. For example, failure of an access gateway can cause the tear-down of all call routed through that access gateway at the time. If the access gateway can serve up to 336 simultaneous calls, then up to 336 calls could be lost due to equipment failure.

We expect these kinds of failures to be rare events, and we are designing the voice network for 99.999% uptime.

In case of a fiber cut, the voice network recovery strategy will be different on our leased network than the recovery strategy will be on the built network. When a fiber cut occurs on the leased network, routers on the voice virtual private network must re-compute their routing tables, which can take up to 7 seconds. Calls in progress on the cut links could be dropped, but retried calls will succeed after 7 seconds.

On our built network we will be using SDH rings, which can automatically recover from a fiber cut within 2 seconds. In case of a fiber cut on a SDH ring, calls in progress across the cut line will experience a 2 second period silence after which the call will resume as usual. This should be contrasted with circuit switched networks, which usually drop calls in progress in the case of a fiber cut.

Networks

If poor quality is not an intrinsic characteristic of VoIP as such, the perception of poor voice quality over the existing public Internet must flow from the lack of QoS supporting technology in that network. As we have indicated earlier, next generation networks like the one Level 3 is building, are explicitly designed to introduce quality of service and reliability in an IP environment, witness the above described results of the quality of service measurement performed by Level 3.

In our view, at least the following elements are critical in order to achieve quality of service in the voice network :

As explained in section one, this quality oriented design and the resulting voice parity with the PSTN, are the distinguishing characteristic between the existing Internet networks and the next generation networks. It justifies a shift in perception of these networks and the services provided on top of it, in the sense that it is appropriate to equate them now to basic, carrier-grade, public networks. Further, it has been explained why these NGNs are in the long term likely to replace the PSTN. It is, however, important not to adopt a simplified view on this evolution :

The foregoing reflects Level 3’s fundamental view on the emergence of VoIP and, as will be further developed in chapter 3, influences our regulatory position towards this evolution. Level 3 has paid full attention to the further questions raised by the ART, notably in sections II and III, but, at this stage, would prefer not to take position on more detailed issues like pricing structure or settlement mechanisms between various players.

Chapter three – regulatory observations

In chapter two we have asserted that VoIP is rapidly evolving, in terms of quality and functionality, from a hobbyist nicety into a carrier-grade public service. We have also asserted that this changed product positioning has fundamental regulatory consequences, in that there is no longer ground to treat VoIP services differently from their circuit switched counterparts. As ART itself indicates, the legal definitions of voice are " based on practical service models, which may be considered to be technology-independent ". This means, in essence, that there should be no distinct regime for IP based voice services. It does not mean, however, that we consider the existing common regime for public networks and voice services as a perfect one.

Legal issues concerning Internet Telephony

As we explained repeatedly, we do not think it is useful to define Internet Telephony in terms of underlying technology, architecture, let alone a coincidental combination of types of terminal equipment. Academic research may have an interest in developing typologies and the like, but from both a policy and strategic point of view, what counts is what counts for the customer, and this, indeniably, refers to quality and functionality, and in this respect there is no single reason to treat VoIP differently from the diverse inhabitants of the circuit switched voice services universe.

Therefore, the one and only approach to make sense is a definition in relation to the legal, technology-independent definition of voice. In this context, we have, at this stage, no substantial problem with the European Commission’s notice on the " Status of voice communications on Internet ". Since the publication of this notice, VoIP has evolved so far that it satisfies now all relevant criteria to be considered as public voice in the sense of the Services Directive :

Commercial offer

The legal definition of voice implies that the transport of voice is provided as a separate commercial activity with the intention of making a profit. This is indeed what Level 3, and many others, are doing, or intending to do. However, this requirement may in the medium to long term have to be revisited as convergence and network integration materialise in more advanced product concepts, where customers no longer buy for instance voice service, but merely network connection at specified bit rates. This is when convergence will have reached the point where it completely blurs traditional regulatory concepts and the existing definition of voice is no longer relevant.

For the public

The IP voice services offered by Level 3 and others are available to all members of the public, subject to them entering into an arrangement governing the commercial terms and conditions.

To and from public switched network termination points

First, from a call origination point of view, this will generally be the case, as for instance Level 3 is constructing fiber networks or otherwise providing its customers with direct and indirect access to its network platform. However, this criteria risks more than the others to attract technology-dependent interpretation, for instance by referring to termination points on the PSTN, which generally means the circuit switched networks supporting POTS. Moreover, the criteria also risks to attract interpretation which is flawed by its reference to an outdated regulatory framework, namely one where incumbents enjoyed a monopoly right with regard to the PSTN. We have seen assertions that termination points on new entrants’ facilities-based networks are not PSTN termination points. Obviously, this makes a caricature of any regulatory framework for network competition.

Second, from a call termination point of view, this criteria brings us back to the question of interconnectivity. As the Commission indicates, this implies " enabling any user to communicate with another termination point in the sense of any-to-any ". In our view, any-to-any communication implies the capability to place outgoing calls to, as well as receive incoming calls from any other telephone service subscriber in the world, be the destination number located in the same geographical area or the subscriber connected to a different network. In short, for VoIP to be considered as public voice in the sense of the Services Directive, this criteria requires IP-based voice networks to interface with the PSTN’s SS7 platform in order to guarantee seamless any-to-any interconnectivity with the PSTN.

Direct transport and switching of speech in real time

At the time of publication of the Commission’s notice, it was considered that the majority of then existing Internet telephony applications did not satisfy the real time criteria. The Services Directive is not of great help in trying to get grasp of the concept of real time. Comments in the Commission’s notice like " voice mail does not occur in real time" may be enlightening but that will be it. The notice further assumes, quite correctly at the time, that " the time period required for processing and transmission from one termination point to the other is generally still such that it cannot be considered as of the same quality as a standard real time service ". To comments in this respect :

The latter comment should not be taken out of context. It is in no way meant to downplay the critical importance of quality of service. Level 3 is fully committed to achieving the highest standards and is in favor of imposing quality of service standards as a license condition, because of its implications for end-to-end quality in an interconnection environment. Without quality guarantees attached to a license, the right to interconnect which goes with the license would in effect undermine the quality of the public voice service.

Our standpoint in relation to ART’s other questions follows from this analysis.

Licenses

It is Level 3’s opinion that, under existing legislation, a license obligation rests on anyone who intends to provide real time voice services to the public on an any-to-any basis. Although we did not address the issue of value chain here, the addressee of the obligation will be the person or legal entity who controls the equipment necessary to operate a voice service.

VoIP operators, much the same as voice over circuit operators, may or may not use new infrastructure, deployed by themselves or otherwise leased on a wholesale basis from someone else. This is only relevant to the extent it may affect the achievable quality of service and thus compliance with the quality of service condition in the voice operator’s license. As we have explained in chapter two, this risk is real in an IP environment, which is the reason why NGN operators are constructing their own, new networks. From a regulatory standpoint, it is clear that these networks qualify as public networks, since they provide the platform for a public voice service.

As stated before, underlying these viewpoints is the observation that VoIP has grown into a carrier-grade service which is capable of providing PSTN quality and functionality in a transparent way. Transparency here refers to customer interface in the first place, but it also refers to the fact that the Level 3 network support SS7 based interconnection interfaces. Finally, transparency implies that the Level 3 network generally supports regulatory requirements like call monitoring, quality of service and the like. This also implies that Level 3 sees no major problems in implementing the license conditions which are traditionally imposed on voice over circuit operators.

However, this conclusion should be regarded as provisional, since it is not unconceivable that technology-related obstacles are encountered later on in the product development process. Moreover, it does not necessarily mean that Level 3 thinks the current regulatory regime is perfect.

International context

Our analysis here is based on the Commission notice which applies throughout Europe. Insofar as the described technology and market evolution takes place in other European countries, it can be expected that regulators will evolve in the same direction and equate VoIP to voice over circuit for regulatory purposes.

On a more global level, whether Europeans are capable of influencing the implementations of an international framework with regard to Internet telephony, and, more specifically, whether the EU can influence the US approach of minimum Internet regulation, remains to be seen. However, it also remains to be seen in what way this may be desirable. Although in the US the tendency to treat Internet telephony as " enhanced " is particularly strong, be it only to escape access charges, we believe that a similar shift in perception will take place there and that the US will evolve towards incorporating VoIP into the common carrier regime which now applies to traditional voice operators.

Numbering

Our views on numbering follow from the basic assertion throughout this document that VoIP has evolved into a carrier-grade voice service and, as a result, should be incorporated in the existing regulatory framework governing public voice :

Furthermore, as VoIP is entirely transparent to end users, there is no objective ground to introduce a separate regime for the publication of VoIP subscriber numbers.

Interconnection and interfaces

Although this may change as VoIP networks grow, the overwhelming majority of interconnection traffic to and from carrier-grade VoIP networks will be with the PSTN. This typically includes

The interface with the PSTN is entirely transparent because not only the VoIP network supports the SS7 standard protocol but also IP/PCM conversion takes place in the voice gateway. In other words, to the incumbent, VoIP appears as traditional voice delivered on E.1 interconnect channels in an SS7 environment. One consequence is that traditional interconnect charging principles continue to apply.

Other regulatory issues

ART raises some other interesting regulatory issues, but it is our opinion that the underlying supporting voice technology is not relevant in the discussion of these. For instance, we may have substantial comments on the universal service regime, but these will not stem from the fact that we use IP in our voice network. Or we may feel that the provisions with regard to emergency calls are too vague, or not efficient, and that it is high time to review these provisions together with ART, but such course of action has no connection with underlying technology : as Level 3 is capable of supporting full transparent PSTN interconnection, the provision of free access to emergency services is no issue with regard to those service configuration where such provision makes sense. Of course, technology may come into play where regulatory requirements necessitate specific network internal arrangements, which is is obviously the case for call interception, but there is no obstacle to supporting those requirements transparently.

Conclusion

Rather than dealing with all questions in detail, Level 3 has attempted to substantiate the fundamental axiom that VoIP has reached such degree of maturity that it is to be equated to carrier-grade public voice, both from a customer perception and a regulatory point of view. We are convinced that insertion in the existing regulatory framework is the only sensible approach in this context, if the public policy purposes which underly this framework are to survive the arrival of the next generation network.

It is nowhere implied that the regulatory framework is perfect, far from there. The point Level 3 has been making throughout is that the necessary review of the regulatory framework should focus solely on the nature of the services provided and not on the technology supporting those services. To keep on isolating VoIP from the common regulatory framework for voice services would be tantamount to regulating against the unavoidable evolution of the industry.

As this issue is at the very basis of Level 3’s existence, we remain committed to an open and continuing debate with ART, as well as with the wider forum of regulators and market players who share our interest.

Jo Van Gorp
Bruno Vanneuville
Level 3 International
©Autorité de régulation des télécommunications - Septembre 1999
7, Square Max Hymans - 75730 PARIS Cedex 15
Téléphone : +33 1 40 47 70 00 - Télécopie : +33 1 40 47 71 98