WAVE Report

NGN 2002
by John Latta
Wave Issue 0239 11/1/02

Boston, MA
October 15 - 17, 2002

NGN 2002 continues to offer the stark reality of the perils of the NGN economics and business potential. John McQuillan, provided his frank assessment of the market and the long term implications. Bottom line - there is no light at the end of the tunnel.

The telecom collapse numbers are staggering.

600,000 jobs lost (only 350,000 were gained due to the bubble)
11 of the top 25 money losing firms in 2001 were in telecom and $107B was lost;
Telecommunications is under a huge $470B mountain of debt;
61 Service Providers are in bankruptcy and this is continuing with an average of 1 per week going bankrupt;

The market dynamics are unrelenting:

Declining prices for voice;
Declining number of phone lines;
Declining prices/bit/sec;
Shift from phone and fax to e-mail;
Shift from wired to wireless;
Shift from TDM to IP.

One of the most profound statements John made was that

"Technology has created an enormous deflationary pressure."

That is, the success of technology has undermined the industry that created and deployed the technology. In the case of the telecommunications industry, both optics and IP, were so successful with the attendant lower cost business models that these technologies were responsible for the havoc that has been caused. The revenue curve did not keep up with the economies of scale that the technology brought.

The PC industry has many of the same effects but not as drastic. Intel has spoken of the software spiral. The demands for computing resources have been fueled by the demands of software which needed more MIPs. This spiral chain has been broken. The drive to buy the faster clock speed microprocessors is no longer driven by the software performance imperative. Thus, for Intel to flood the market with ever faster clock speed microprocessors, that leverage Moore's law, only hastens deflationary pressures on the PC industry. One is left with the nagging question - is the deflationary pressure on the PC industry only building to ultimately get worse?

The financial markets are no better off. There have been 60,000 jobs lost in investment backing since January 2001 and no acquisitions or IPOs in this sector since then. Likewise, in the sector there were 2,500 companies taken public from 1996 - 2000. There are now 10,000 venture backed private companies that have no exit strategy. It is estimated that at least 7,500 and possibly up to 9,000 could run out of money between 2002 and 2005. No matter how good the technology, the markets are providing no way to exit and to secure public financing. A massive body of intellectual capital is and will continue to be lost.

At the last NGN the ILECs were largely immune to these forces. No more. They have lost half of their value since 2001.

The discussion by John also included ways to characterize the current environment. Some call it the "Perfect Storm," others the Neutron Winter, and others the Sorceror's Apprentice. John expressed it best by characterizing the time from 1999 - 2000 as being in the center of the storm and now we are in the clean up phase of what is left. It was his view that the market conditions will get worse before they get better and we will not go back to the good times of the storm.

Other comments made:

Money, politics or a monopoly trumps great service provider technology any day.

ENUM, the electronic numbering standard, is a great move forward to having the universal access number.

In spite of the significant technology developments at the core, i.e., fiber, there is still no broadband to the home.

In the Plenary Session, entitled Broadband Deployment Challenges: Stimulating Supply and Demand, President Bush's team was represented by Michael Gallagher, Deputy Assistant Secretary, National Telecommunications and Information Administration (NTIA), US Department of Commerce. The panel had many difficult issues to discuss including:

A National Broadband Policy;
The role of Open Access;
How much bandwidth is necessary for residential customers; and
What are the impacts of walled gardens?

When John Passmore asked Michael Gallagher about the Administration's view of the impact of deep packet analysis, which could be used to detect IPSec usage, and thus ILECs shutting down customer sites, Michael seemed surprised and did not understand the issue. This is at the heart of the topic raised at VON - what will the government do to support an open end-to-end Internet? So far, we have not gotten a good answer. One of the stock responses is that "competition" will fix everything. As many in the audience have stated - there is not effective competition at the local level in the delivery of broadband. Maybe one of these days the Bush Administration will get its head out of the sand on the practical realities of broadband delivery.


AT&T's CTO, Hossein Eslambolchi gave the keynote. Points made include:

"IP will eat everything."

E-mail may be the killer app but it only takes .7% of the bandwidth on AT&T's backbone while HTTP/HTTPS only takes from 16% - 17%.

If we are to look for compelling applications of the Internet we need to also look at those that use bandwidth.

14m dial-up AOL users take the same bandwidth that 1m AOL broadband users do.

While the PC is rebooted to recover from issues, a router in the network can never be rebooted.

At AT&T we have 70 billing systems.

A key initiative at AT&T is the concept of Zero. This is to drastically reduce the number of systems which AT&T must manage and operate.

The only place where AT&T can differentiate is at the edge of the network.

Presented was a concept of SOIP - Services over IP. Applications are services. (AT&T is beginning to get it.)


Christian Hutiema, gave both a blunt and informative talk on what Microsoft is doing to accomplish the transition to IPv6. Points made include:

NATs are evil. For IPSec packets, crossing one NAT is difficult and 2 NATs is impossible.

The transition to IPv6 is largely driven by the need for more address space, however, it also provides for true mobility and at the same time improved security.

The tools to accomplish the transition include:

IPv6 stateless address auto-configuration
6to4: Automatic tunneling of UPv6 over UPv4
Automatic tunneling of IPv6 over UDP/IPv4
ISATAP: Automatic tunneling of IPv6 over IPv4 for use behind a firewall.

Laid out a three phase approach to handle IPv6 transition in

Home Network; and

The 3 phase approaches made a lot of sense. Christian stated these approaches are close to being corporate policy for the transition.


This is a standards body working on the transition to IPv6 in the cellular phone network. They expect to have 1B phones that use IPv6 and regard the transition as both critical and mandatory. It was also interesting to note that all these phones will be SIP based.

Quotes from the day:

The ILECs have from 60 - 100 lobbyists on Capitol Hill while the only remaining CLEC, Covad, has 2.

One of the major values of SIP is that is separates the control plane, which is basically what SIP does as a signaling protocol, from the media. Thus, one can get control information on a PDA about a phone call if the phone is busy.

The enterprise phone market is about $30B per year. In 2001 the enterprise IP Phone market was $1B and is expected to grow to $2B - $3B by 2002. Currently, this implies that IP telephony for the enterprise is about 10% of the market.

IP telephony eliminates distance as an issue in communications.

DSL back haul issues have become significant for the ILECs. A comparison has shown that the ILECs pay $5 - $10 per subscriber over cable in accomplishing backhaul. Note that cable has this built into its plant.

Paul Nikolich, Chairman, IEEE 802 LAN/MAN Standards

Paul had much to say:

Under the 802.11 Next Generation study group, mesh networks are being considered for standardization.

An 802.16 Study Group has an activity to examine how fixed wireless, i.e., LMDS and MMDS, can be made to support mobil.

There is also a study to examine how a mobil service, PCS, can be made to support fixed wireless.

It is expected that new security requirements for 802.11 will drive up the costs of access points to meet these needs. Security will include control messaging for user and network authentication. New WLAN access methods to support security will include:

RADIUS server-based authentication;
Access points configured as RADIUS clients;
User accounts must be created and shared between cellular and WLAN cells; and
Certificate infrastructure required to issue user and computer certificates.

This is intended to allow for seamless roaming between WLANs and 3G networks without compromise of security.

A critical issue lies with portable 3G devices - their high power consumption - and thus short battery life.

There is an effort, called project Rainbow, a joint endeavor by Intel, IBM, AT&T Wireless, Verizon Communications and Cingular Wireless, to create a nationwide Wi-Fi network.

Paul closed by predicting that 3G technologies will ultimately be replaced by an IEEE 802 mobil broadband wireless solution.

WAVE Comments

The scope of the impact of these activities could be enormous. Both LMDS and MMDS largely lie dormant due to the inability of CLECs to make them commercially viable in fixed wireless applications. The prospect that this could be converted to mobil high bandwidth services is very important. Further, the notion of a Next Generation 802.11 set of standards is also important. The prospect of Mesh Networks as a part of the 802.11 implement is also significant.

Watch for developments in these standards efforts which will impact products for years to come.

Paul's comments that 802 based standards would replace 3G is a gutsy prediction but one that Mesh Networks is seeking to implement.

Keynote - Terence Matthews, Chairman, Mitel

Much of the talk was focused on his start-up of Mitel in the 1960s and his recent buy back of the firm with personal money. He is now Chairman. Comments Terence made include:

The capabilities of fiber are illustrated with the Apollo undersea cable recently laid from the US and Europe. It supports 3.3 Tb/s--30% more than all the existing undersea cables from the US and Europe.

One application of bandwidth is an always-on video wall. This enables the creation of shared spaces and a virtual presence. It has significant applications in vertical markets and telemedicine is only one.

There are 90m SME's worldwide and 8m in the US. The major challenge that lies ahead is how to service these companies. They have no IT staff and the focus is on only how to do business not on computers or telecommunications. 50m of these SMEs would like to look bigger than they are via the Internet. We are working to accomplish this. One element in our plan is the 3500 phone system which is software and SIP based.

We need to focus, in this industry, on "Enhancing the Quality of Life" using the digital infrastructure now available. (WAVE Comment - Totally agree)

Merwyn Andrade, Aruba Networks

The talk was entitled "Wireless LANs: Coming out of the Closets." Merwyn's focus was on the changes needed in 802.11 to support corporate WLANs. Further, he also highlighted the progress in various 802.11 standards. His points included:

Ratification and product availability for 802.11g and h would happen by the end of 2002. However, 802.11h interoperability testing would begin in Q1 2003 while the same testing for 802.11g would not happen until Q2 or Q3.

The corporate needs for WLANs will cause important changes in the implementation and deployment of the technology:

Expect to see WLAN switches by 2003;
Much lower cost APs;
Design of WLAN components for minimization of "plenum rolls."
A scalable control plane;
Access to RF visibility as a part of network management; and
Intelligent wireless access ports;

The sense was that Aruba Networks has such products but they were not described.

Rick Rotondo, Mesh Networks

Two problems with 802.11 are that it was not designed for mobility and the networks are capable of self-interference. If these two issues can be addressed, 802.11 is in a position to seriously challenge 3G. In short, to compete in the market WLANs need to grow up. By implementing a mobil meshed backhaul network it is possible to reduce the number of access points and enable mobility. Mesh Networks has a QDMA technology, which is similar to CDMA, which allows for 802.11 mobility. This has been demonstrated on a bus trial in Florida.

In our testing of 802.11b the latency between hops was only 1ms. Thus, the network, even with nodes, can support VoIP. However, the proposals for QoS may add more latency.

In the design of these networks there is an "iron triangle" made up of Throughput, Distance and Power. Compromise one and there must be a compensating change in one of the other parameters. Raise thruput and either the distance must go down or the power up.

I spoke with Rick and he made the following interesting comments.

802.11g is a dark horse. 802.11a is already moving to be deployed and 802.11g, when it comes out, will have to contend with 802.11b. The prices of 802.11a are falling so fast that 802.11g cannot make a cost advantage argument.

Mesh Networks is moving ahead to get its technology adopted by the NIC card vendors. However, this effort has shifted to the silicon companies to get the technology buried into the silicon.

IETF Update

These are notes from all the presentations in this session.

A major update to the SIP RFC was released with RFC 3261.

SIP now supports both IM and Presence.

SIP is very good at rendezvous - that is finding the call termination.

SIP is capable of integrating with cell phones and even IM chat rooms;

A new feature within SIP is the capability to support pseudonymous calls and location services.

ENUM is moving very quickly. This is a way to map an internet address to a phone number, which both the PSTN and SIP can use. URI's translate to phone number equivalents.

One of the major advantages of VoIP and ENUM is the ability to drive costs out of inter-enterprise calling. That is, it eliminates the toll call cost between two geographically disperse locations in an enterprise. ENUM takes this saving a step further and enables those outside the enterprise to have the same benefits - lower cost calling.

Recently the US DoC received a letter from all the ILECs and many others requesting an early deployment of ENUM. As a result there are similar adoption actions overseas. Phones capable of ENUM calling are expected in 6 months.

I asked the question: "Does the ITEF have a license free CODEC so that low cost SIP phones can be made?" Yes, we have a proposal for such a CODEC in front of us but do not have the skills to evaluate it.

Notes from the talks:

KDDI, the second largest cellular operator, was able to sell 2.6m phones in the first 6 months of operations of its CDMA 2000 based 3G infrastructure. NTT DoCoMo, in contrast, has had considerable difficulty with its roll out of W-CDMA. One reason for the KDDI success could be the camera built into cell phones - they sold 900,000 phones as a part of the 2.6m sold.

It is expected that the relationship between WLAN operators and equipment providers could be very similar to the cell phone companies and operators. That is, WLAN devices will be made to exercise the capabilities and applications on the WLAN, and it is the operators who will push these devices in the market.

A major cost in cellular implementation is the base station. A base station which covers about 5 miles will cost up to $100,000 while a picocell, with 1,000' radius will cost $20,000.

A major OpEx cost of cellular is the back haul from the cell to the switch. This same cost constraint is applied to access points in a traditional WLAN environment. However, with a mesh topology the dynamics may be quite different.

ZigBee Rolls Out

It was hard to believe but another home networking standard surfaced today. This is focused on control and is being done under the IEEE 802.15. Pat Kinney, Board Member, of the ZigBee Alliance gave the presentation.

An enduring question on the justification for ZigBee is - Why not Bluetooth? In response, ZigBee claims:

Modulaton DSSS FHSS
  11 chips/s 1600 hops/s
  62.5 K symbols/s 1M symbols/s
  4 b/symbol  
Peak Information Rate 128kb/s 108 - 723kb/s
Battery Life 2+ Years Mobil Power - charging - 1 day typical
  Designed to optimize slave power use Designed for ad-hoc use

Other features of ZigBee include:

Data rates of: 250kb/s, 40kb/s and 20kb/s
Topologies: star, peer-to-peer and mesh
255 devices/network
CSMA-CA Channel access
Optional guaranteed time slot
Fully handshaked protocol
Dual PHY (2.4GHz and 868/915MHz)
Range: 10m nominal but can go from 1 - 100m
Location aware: Yes, optional

Members of the Alliance include.

Promoters: Honeywell, Invensys, Motorola, Mitsubishi, Philips
Participants: Adcon RF Technology BV, AlfaPlus Semiconductor, Inc., Analog Devices, Atmel Corporation, Cambridge Consultants, Chipcon, CompXs, CSEM SA, Eaton Corporation, Figure 8 Wireless, LLC , France Telecom, Intel Corporation, Leviton Mfg. Company, Inc., Microchip Technology, NTRU Cryptosystems, Inc., RF Micro Devices, Xanboo Inc., Xemics,


SIP Session

This event was a SIP praise event. It was hard to believe how quickly the landscape has changed. Last year SIP barely hit the radar screen. There are interesting parallels between MPLS and SIP. MPLS has brought to the backbone of the Internet connection capabilities based on labels for routing. As a result, at least on the backbone, the Internet can effectively look like ATM. This impacts traffic engineering and more importantly VPNs which look like TDM circuits. SIP goes one step further and now brings an application to the Internet - voice. Using the open standards approach of the IETF, SIP has risen rapidly to enable telephony with a simple text based signaling protocol. As a result H.323 is on the skids. Thus, the Internet can be an emulator of the PSTN and certainly ENUM goes a long way to making this a practical reality. Another reality of the use of the Internet is rise of free on-network phone calls from SIP based service providers. This matches the same type of service on Yahoo BB.

One disconnect in comments was that

MPLS did not need to extend to user devices and
MPLS does not have an interoperability issue.

That is, most backbone companies with MPLS are seeking the edge business where the labels are attached and be the same companies that also terminate the pipe at the other edge. One should not take these as givens however. The MPLS Forum meetings are a better place to see directions on the use and interoperability of MPLS. There was also an interoperability demonstration at NGN by the MPLS Forum, which I did not tour.

Conference Assessment

NGN continues to be an excellent conference. It offers, in a short span of time, a broad overview of the state of the industry. This is not just a technical assessment but also a very practical one which begins with John McQuillan's blunt appraisal of the dismal state of the industry.