Features

The Continuing Evolution of the Wireless LAN - Hotspots and the Banias Effect

by Nick Hunn | posted on 12 January 2003


Home wireless LAN is advancing by leaps and bounds. Unless the industry can resolve its internal issues over standards creep and security problems, the home market will exceed the business market by the end of 2003. Intel's Banias chip -now officially "Centrino" could change the ownership of wireless LAN. With the hardware becoming endemic in the next 18 months, the timing is perfect to allow the mobile network operators to wrest control of its usage from the PC industry.

"Although mobile operators are best placed to win the hotspot market, if they fail to grasp the opportunity Microsoft could launch the ultimate solution by adding hotspot access as an MSN service"

Evolution of the Wireless LAN – Update Winter 2002/3

Earlier this year I reported on the conflict between the different wireless LAN standards and my belief that 802.11b would come out as the winner, based on usability and cost effectiveness, rather than necessarily on raw throughput. The last six months has seen some dramatic developments within the wireless LAN industry in terms of silicon and business models, and the time is appropriate to return to the subject.

Very little of the underlying user requirements have changed in those six months. Wireless LAN is steadily advancing, particularly in the U.S. and Japan. However, the headline shipment numbers are increasingly obscuring the demographics of where the real deployment is happening. Analysts, such as Yankee Group, Dell ‘Oro, In-Stat and Synergy have indicated that the increase in shipments has slowed or is tailing off from the exponential projections of 2001. A deeper analysis shows that although the overall numbers remain high, corporate implementation is in fact falling and the major growth has been in domestic installations.

<1/> Effect of SoHo Market on Wi-Fi

Two main reasons can be advanced to explain this trend – security and standards. The security concerns around Wi-Fi have started a major dip in corporate acceptance which will not be rectified until the end of 2003.

In both of these areas – security and standards, the standards body and industry voices have been vying to shoot themselves in the foot and discourage corporate take-up of wireless LAN. Unfortunately there seems to be little appreciation of either of these facts by their perpetrators and the behaviour of the industry remains one of alienating what they see as their core user base – the wireless office. The relative success of the consumer market is largely because these users have a more realistic approach to the technology of "Does it fulfil my need?"; security and long term infrastructure investment are problems that only get addressed in the home when the product falls over.

Wireless in the home is compulsive. I have yet to meet anyone who has installed and used wireless at home who would willingly give it up. Perhaps even more so than wireless in the office it changes internet usage patterns and social interaction. Once a wireless laptop is available at home your approach to computing changes.

With wired access, computing occurs where there is a physical connection. This invariably means that it becomes a solitary, anti-social activity. The availability of a wireless laptop brings the family group back together within a single room and establishes the laptop as a shared resource for internet access, in much the same way as the TV or hi-fi is used. Add to this the fact that the laptop is probably on standby, and the instant access to the Internet changes the usage model from a session based one (on the grounds that by the time you've booted the desktop, got Windows running and connected via your modem you might as well spend half and hour on it) to a far more "on-demand" approach, where a specific piece of information is found and the laptop is then put down again.

Wireless in the home lets you "graze" information as one of a family group.

It is interesting that this is exactly the paradigm of "Always-on" that is touted as the dominant usage model by 3G proponents. Its compelling nature at home suggests that once this behaviour model is established, then it can be extended to Wi-Fi hotspots and GPRS Wide Area access provided the pricing model is acceptable. That, as we shall see, is a very big proviso. If the current purchase and implementation trends continue, by the end of 2003 there will be more home users of wireless LAN worldwide than office based users. A large proportion of these may justify the initial home installation for business use, but the fact remains that actual usage will be led by personal installations and not by corporate policy.

To change this trend the underlying problems of security and standards creep must be addressed.

<1/> Users of Wi-Fi products

Security

Wireless LAN's are insecure. There are many different ways of quantifying the degree of insecurity, but on a business level of security the touchstone is the wired network within the four brick walls and firewall of the corporation. Any network manager who values their job is going to be loath to install a technology that lowers that existing level of security.

Having said which, any technology that allows external access to the business will threaten to breach that security. This is equally true for modem and ISDN, but frequently ignored. A few examples illustrate the problem.

It was recently reported that the LED's on most modems are connected directly to the data transfer stream. If you have a spare Friday afternoon get a toy telescope and photodiode, focus through the office window on your modem and can scare yourself at how easy it is to hack into your data transmission. It is also instructive to consider how secure the telephone switchboard or LAN is in the hotel where you're connecting? How much would a competitor need to pay the receptionist to tap into it? The reality is that none of the public networks are secure (with the possible exception of Bluetooth and GSM).

Anyone worried by security (and that means almost every business customer) must ensure that their data is encrypted at the physical link. That doesn't preclude the use of these networks – you just need to take the appropriate precautions, which generally mean a Virtual Private Network (VPN).

Where Wi-Fi made a mistake was trying to deny this basic truth. The standard made a good attempt at incorporating elementary security with Wired Equivalent Privacy (WEP), but it must be understood that it is not equivalent to a true cabled connection. From the beginning this should have been presented as a starting point for a secure solution. Instead, when WEP was cracked, the barriers went up and rather than admitting the truth the industry did its best to rubbish those criticising it. The inevitable result was to encourage still more people to find ways to crack it. We were then promised cake tomorrow with a new improved security solution, and sure enough that cake got cracked too. Telling the truth would have placated the corporate network managers and let them get on with a holistic approach to integrating wireless LAN. Instead we now have a situation where every executive within the corporation has heard that wireless LAN is insecure and who will question the safety of installing it.

This reaction of the wireless LAN community is reminiscent of what happened when companies first came across the problem of their garbage cans being rifled for confidential documents – their immediate reaction was to put the documents into black plastic bin sacks. It was only after these were repeatedly stolen that it dawned on them that the real solution was to destroy the documents first by shredding them. Until the wireless LAN vendors and standards bodies progress past the black plastic bin bag stage I have grave reservations about the corporate uptake of Wi-Fi. Simply introducing new, improved, go-faster security "bolt-ons" at this stage is probably more a challenge to the Wi-Fi hackers rather than a reassurance to the totally sceptical potential user. There are a limited number of times that you can cry wolf. It would be better to start imparting a consistent message about how to install Wi-Fi as part of an overall secure system.

Standards

I've covered the insane promulgation of 802.11 alphabet soup in a previous report. I had hoped that the passage of time might have bought vendors to their senses such that they concentrate on 802.11b and 802.11a, but the folly continues. It seems that the industry has totally forgotten the customer and their requirements, and I reiterate again the fundamental that must be understood:

It has taken the corporate ten years to migrate from 10 Mbps to 100Mbps wired LAN at the desktop, and will take another ten years to introduce 1Gbps. How can the wireless LAN industry expect the corporate customer to upgrade their wireless infrastructure every eighteen months?

Despite every cry for common sense the industry is still rushing towards 802.11g – a standard that is both unnecessary and unlikely to deliver any of its promise in the real world. Meanwhile customers have shown a total lack of interest in 802.11a, which is hardly surprising as there are precious few wireless users who have advanced to the stage where they need more speed, but still we see constant pressure to improve and enhance the standards rather than concentrating on growing a stable critical mass of wireless.

What wireless LAN needs is stability to attain that critical mass. But the prospect of installing infrastructure for a constantly moving standard will guarantee corporates delay their installation plans. In 2001 they were being told to install 802.11b, in the first half of 2002 it was 802.11a, then 802.11g and now they're being advised to hang on as 802.11a+b is the killer solution, just around the corner. Is it surprising that they are voting with their feet for cable drums of Cat 5?

<1/> What flavour of wireless LAN today, Sir?

The Banias Effect

Chipset costs for 802.11b have been falling dramatically over the course of the year. Almost all of the 802.11b silicon shipping is still from the three core vendors – Intersil, Agere (formerly Lucent) and Atmel, but these have been under pressure to reduce costs from a raft of new entrants. The announcement of 802.11a+b solutions from Atheros, Envara, IceFyre and Synad amongst others has also added to the pressure to cost reduce 802.11b.

By the middle of 2003 we expect that the silicon cost of an 802.11b solution will have fallen to around the $12 mark, with a total Bill of Materials cost below $20. Although that is immensely significant in driving adoption by laptop manufacturers, it will be dwarfed by the effect of Intel's new Centrino - formerly, Banias - processor.

Intel has widely announced its next Centrino processor chip, aimed at the mobile market, and referred to in most documents by its project name, Banias. This is primarily intended to solve many of the power problems of previous mobile processors; but it has a massive significance for wireless LAN. It is significant because the 2002/2003 chipset contains support for the 802.11a and 802.11b wireless LAN standards, requiring only the addition of an external transceiver and modem interface. The effect of this is that a laptop manufacturer who uses the Banias chipset will have a Wi-Fi "ready" motherboard.

Intel has promised to provide the transceiver in the form of a mini-PCI interface using their own silicon, codenamed Calexico. They have hinted at prices of $20 for an 802.11b solution and $35 for a combined 802.11a + b product. We believe that this figure is unnecessarily high. An analysis of the components that are expected to be required to turn Banias into a full WLAN solution suggests that a manufacturer wishing to put the transceiver onto the motherboard could achieve it for little under $9 for 802.11b and $22 for 802.11a+b. At this price point it seems likely that a large percentage of Banias based laptops will incorporate 802.11b.

Although Intel is pushing the 802.11a+b combination, I do not believe that many manufacturers will jump at the chance of incorporating it. The limited availability of 802.11a products has delayed a number of fundamental questions being asked about its global acceptance. There is still a lack of worldwide approvals for 802.11a, with very limited acceptance within Europe. In addition, whereas 802.11b is becoming deemed safe by airlines, the operating frequencies of 802.11a may well cause them concern, as they lie much closer to the frequencies where several radar systems operate. Experience shows that users will not understand that they need to turn their Wi-Fi adaptors off when in the air (it is difficult enough to convince them about turning mobile phones off). So until and unless the FAA gives 5.1GHz a clean bill of health, incorporating 802.11a could reignite the hoary old chestnut of banning laptops on flights. That prospect, plus the additional RF component costs will probably ensure that most vendors opt for 802.11b only as the primary option with Banias.

<1/> Anticipated Implementation Cost for Wi-Fi on the motherboard with Intel's Banias

Banias presents a problem for AMD. Possibly galvanised by earlier Banias announcements, they have recently launched their Am1772 Wi-Fi chipset which provides a "soft" baseband with a PCI interface. Although this goes some way towards providing wireless support, it is less integrated than the Banias approach and will almost certainly carry an additional implementation cost of around $6–7 over the Intel solution.

This is good news for the laptop purchaser, who will see the cost of adding Wi-Fi to a laptop falling significantly during 2003. At the moment the cost of a PC Card is still around $75. As integrated Wi-Fi appears the price differential the customer pays for enabling the laptop will probably drop to below $20, with Wi-Fi appearing as standard by 2004.

<1/> Perceived Cost of Wi-Fi

Implications for Silicon Vendors

Although the customer will benefit, the introduction of Banias poses a very serious threat to the raft of silicon companies who have invested in wireless LAN. The fact that both Intel and AMD have a low cost, native solution to Wi-Fi on the motherboard excludes these vendors from the market for new laptops. Instead they will be forced to concentrate on the legacy laptop market, (where price competition will be fierce), the desktop USB market, the PDA market (which is small and power sensitive) and the access point manufacturers. These markets will also be attacked by Intel and AMD with "soft" Wi-Fi solutions that remove the baseband processor cost and benefit from the economies of scale both will enjoy from their integrated PC market.

A crucial concern for this legacy market will be the perceived cost of adding wireless to a laptop. It is not yet clear what percentage of laptops will have the full RF solution placed on the motherboard. For those that do not, some manufacturers may provide connectors for a wireless LAN transceiver module to be added. Intel has already identified this, and suggested that it will supply mini-PCI transceivers, with an anticipated market price of $45 for 802.11b and $65 for multimode in June 2003. Once again, the emergence of alternative RF chip suppliers suggests to us that these process will rapidly be undermined by third party offerings. If this is the case, the list price of these transceiver boards may well be advertised below $30, setting a price expectation for the consumer, who is likely to be unaware of the distinction between this and a full PCMCIA card. The inevitable result will be fierce price competition, further commodifying wireless LAN products and ensuring that there is little opportunity for margin either for the silicon supplier or the peripheral manufacturer.

Intel is promising that it will provide RF chipsets to complement the Banias, but its introduction opens up an opportunity for traditional RF chip suppliers to address this particular market. Amongst these, RFMD, Maxim and Micro-Linear have already announced or are shipping RF chipsets which could be adapted to interface to the Banias. Unlike many of the new entrants in the Wireless LAN industry, these are stable, established manufacturers with track records who are already supplying the laptop industry. Where a laptop manufacturer is loath to extend their RF real estate to Intel, these are the companies likely to pick up the business, not the newer start-ups.

The silicon vendors most at greatest risk are those in the 802.11a industry. Intel's approach to wireless LAN will effectively exclude any other of them from the new laptop business. The only motherboard space they can attempt to claim is on AMD platforms, where the current AMD solution does not offer support for 802.11a. (This may be a minor issue, as the corporate customers who would be top of the list to consider 802.11a are almost all purchasers of Intel processors.) However if there is any sign of a take-off of 802.11a AMD will almost certainly pursue its own solution, possibly by utilising a 5GHz radio from one of the established chip companies above, with its own baseband or one licensed or bought from the growing number of wireless IP companies.

The conclusion is that if 802.11a does take off within PCs, the sole beneficiary is likely to be Intel and the indigent, alternative RF silicon suppliers.

The entry of Intel also raises fears for the few specialist vendors who are designing Bluetooth + Wi-Fi chipsets. They will also be excluded from the PC motherboard and will have to stake their claim to survival in the legacy and access point arenas.

Outside the PC, the falling cost of 802.11b wireless routers will ensure that 802.11b dominates the home market. Where only two or three PC's are utilising the bandwidth of broadband into the home the data rates of .11b are more than sufficient. That leaves 802.11a having to convince the corporate market that it is a useful technology to employ within the office (as outside the office 802.11b will dominate). It is difficult to see that this will happen, particularly with 802.11a still illegal within Europe. Unless laptop vendors implement Intel's 802.11a+b solution, then the critical wireless mass will be 802.11b only.

There will be some who find a use for the greater bandwidth of 802.11a, possibly within multimedia applications rather than the PC environment. This may leave room for two silicon vendors to provide a "full" silicon solution targeting legacy and access points, but these will suffer intense price pressure as a result of the price points set by similar 802.11b equipment. Given Intersil's debut in the 802.11a market it must highlight concern regarding the possibility of breaking even for almost all of heavily financed VC start-ups. It is likely to see a frantic scrabble to develop other niche applications for wireless LAN chipsets in areas such as 3G handsets and VoIP phones in an attempt to find applications that are not dominated by the PC and hence by Intel.

Implications for Wi-Fi Usage

In the previous survey I argued that by 2005 most users would have their first experience of wireless LAN outside the office. With the announcement of Intel's Banias chip as the Centrino chip, I'm in the unusual situation of pulling forward the date of that prediction to 2004. The minimal cost of implementing Wireless LAN will ensure that far more users will have wireless enabled laptops than previously envisaged. Intel has already predicted that by the end of 2003 80% of all new laptops will employ the Banias chip and 50% of all new laptops will include wireless LAN. I suspect these figures are high, but still anticipate a rapid growth of Wi-Fi over the coming 24 months.

<1/> Integrated vs add-n Wi-Fi sales

Adapter quantities include current internal Wi-Fi modules, including Mini-PCI format.

By the end of 2003, this means there will be over 24 million laptops out there with wireless LAN searching for places to connect. A year later the Banias effect means that this number will have swollen to 42 million. Hotspot providers need to get their act together to turn these into satisfied mobile users who can provide them with a revenue stream.

<1/> Cumulative Wi-Fi enabled Laptops

Almost every one of these new laptops will be running Windows XP. XP knows about wireless LAN's. If you have a working Wi-Fi adaptor in your laptop, it will pop up a bubble every time it sees a wireless LAN and tell you that "One or more wireless networks are available. To see a list of available networks, click here". Most users do what they're told. If asked to click, they will. And many will then discover that they are connected to a network. It may be a hotspot; it could be an office network or just a private home LAN spilling out down the street. But in a fair number of cases they will find that with no further effort on their part, they have connected to the internet.

<1/> Windows XP knows about wireless

It raises the question of where they will connect. Many will be at home, and the advent of Wi-Fi enabled laptops must help to drive the market for wireless at home.

Today an ADSL wireless router is selling for a retail price around the ¥ 20,000 / £ 100 / $ 150 mark. On e-bay they reach half of that, which is equivalent to a few months ADSL line rental. As soon as the installation difficulties are sorted out to allow a non-technical installation it is likely that this will become a standard option for new ADSL line installation, with the cost of wirelessly enabling the home falling below $ 100 by the end of 2003. Once the ADSL providers observe that laptops are Wi-Fi enabled by default, home wireless will be offered as part of any ADSL deal. The ability to top up the package with USB adaptors for existing desktops will allow them to develop higher priced service offerings which bring broadband to every PC in the home.

It is an important factor for the industry to grasp that this home wireless market for USB desktops adaptors could be as large in terms of total units as the laptop market, yet today most silicon vendors only pay it lip service. There is an open niche for a single chip, CMOS USB 802.11b solution.

With the expected level of deployment in laptops, it seems likely that there will be a pent up demand for public access points. The relative commercial business cases behind these have already been extensively debated, although mostly with a lack of understanding of customer requirement. Nevertheless, a growing number of hotspots are being deployed, alongside a movement of free, community wireless hotspots. The coming growth of wirelessly enabled laptops should be sounding a clarion call to the hotspot industry to prepare themselves for business.

Hotspot business models

After a few years of debate over their worth, hotspots have become the flavour of the month. Every day brings new announcements of partnerships and business plans to put 802.11b into more and more locations, with increasingly inventive business models. From "curse of the mobile operators" hotspots are now seen as indispensable to their success in providing data, although whether this is to the benefit or detriment of 3G usage is still arguable.

Unfortunately few of these hotspots announcements seem to take any great regard of the likely customer base or the user model. Instead they suggest a general feeling that there is a bandwagon to jump on and few are looking before they leap. The remainder of this documents charts the current status of the hotspot business model and how it may develop.

Free Community Hotspots

The initial deployment of hotspots has happened quietly without any commercial backing. Wireless LAN has always attracted free spirited, anarchic elements who see it as a way of bringing broadband wireless access free of charge to a local community. These networks have sprung up across the world where individuals have installed DSL or broadband along with wireless and decided to open up their bandwidth to anyone passing by. The more dedicated and altruistic have enhanced this with the addition of higher power transceivers and antennae. The movement is growing and there is already a cult book on the subject of deploying such networks – "Building Wireless Community Networks", by Rob Flickenger. Although technically illegal in many countries these community networks are flourishing. Whilst many are loose collections of home-based nodes, some are spreading out to give substantial urban coverage. The more organised communities have developed web-sites which provide mapping software to enable users to find the nearest hotspot. To those who belittle this movement, I would particularly recommend visiting the consume.net site to see how much has already been achieved and trying out some of your local access points. It makes the commercial efforts look very immature in comparison.

As broadband comes to more homes with the advent of affordable wireless routers the effective coverage is bound to increase. If commercial offerings fail to work out a sensible charging scheme the vast bulk of wireless LAN users may forsake the commercial infrastructure and instead add to the growth of the free community networks.

The early adopters and cognoscenti are already aware of just how many hotspots exist. Websites already give details of available public and commercial networks around the world. The following table lists some of these sites along with an estimate of the number of hotspots that each covers. There is likely to be overlap between the lists, but it suggests that there are already around 10,000 operational, documented hotspots that can be accessed either for free or by subscription. Those who travel with a wireless card know that there are many more unwittingly providing public access.

<1/> table of hotspots

Commercial Hotspots

One de facto limitation of most of the public sites is that they are generally based in residential areas, as they belong to people who are happy to share their broadband connection. A few companies provide deliberate public access to their wireless networks for visitors and the local community and far more do so unwittingly by failing to set up security, as highlighted by the warchalking and war-driving initiatives. (Secure public access to the internet within companies is a feature that I believe should be actively promoted by access point vendors. It is a courtesy to the visitor to give them internet access, which is sadly forgotten by almost every company I visit. (A recent proposal by one of the mobile networks that this could be chargeable strikes me as being as invidious as the behaviour of a host that charges you for coffee and subsequent use of their toilets.)

This skewed topographical aspect means that open public sites are unlikely to cover the oft-cited business hotspots of hotels, airports, train stations, coffee shops and conference centres, simply because people don't live in these places. These are perceived to be the sites that will address a captive audience and provide the opportunity for revenue. I believe that there is a significant business within hotspots, but to succeed it must consider the needs of the mobile traveller. Very few of the current offerings appear to do this.

The Single Bill

The biggest challenge for the hotspot industry is how to bill the customer. Mobile users want two things from wireless access – a single billing point and a reasonable cost. Today both of these requirements are far from being regularly met. To understand the need for a single bill it's useful to look at a typical business trip and analyse where the user is likely to attempt to access the internet (the assumption being that internet and VPN enables corporate network access and email.)

Day 1

Location: Home Activity: Download latest email before departure 10 mins. Provider: free.

Location: Heathrow. Activity: Wait at airport. Work on mails, browse internet links 60 mins. Provider: Hotspot #1.

Location: Dusseldorf (Office). Activity: Visit Corporate Headquarters – access network during course of working day 5 hours. Provider: Corporate (Free).

Location: Dusseldorf (Airport). Activity: Wait for flight. Check emails 60 min Provider: Hotspot #2.

Location: Copenhagen (Airport). Activity: Wait for train to Malmo. Send emails composed on flight. 15 minutes. Provider: Hotspot #3.

Location: Malmo (Hotel). Activity: Use VoIP to call home for $0.05 per minute. 30 min. Provider: Hotspot #4.

Day 2

Location: Malmo (Hotel).Activity: Check mails before breakfast 15 min. Provider: Hotspot #4

Location: Lund Activity: Visit Company – use internet to display web information 2 hours. Provider: Company network.

Location: Malmo Airport Activity: Check mails. Use VoIP to call office 60 mins. Provider: Hotspot #5.

Location: Stockholm Airport (Arlanda). Activity: Send mails while waiting for train 10 mins Provider: Hotspot #6.

Location: Stockholm Restaurant. Activity: Demonstrate Bluetooth and Wi-Fi to journalist 30 mins Provider: Hotspot #7.

Location: Stockholm Airport (Vasteras). Activity: Update mails while waiting for flight. Call home using VoIP. 60 mins. Provider: Hotspot #8

Location: Home. Activity: Send emails while being shown new website by daughter. 30 mins. Provider: Home (free).

End of trip

Total Time using Wi-Fi Hotspots: 310 minutes. Number of hotspots used: 8

The schedule is a record of a real two day trip, during which I tried to find a Wi-Fi connection in each location. The only ones which worked were at home and in the Malmo hotel, which tried to charge me $15 for each of the two sessions. If every hotspot had existed and taken the same approach, the wireless cost of the two day trip would have been $135. I would also have had the inconvenience of paying eight separate providers and having to document that for my company. It is not a model that any but the most determined user will countenance. Rather than using wireless to aid my business and my productivity it becomes an unwieldy encumbrance which I would rather avoid. And yet I am probably seen by the hotspot proponents as the archetypal target customer.

The key is for service providers to understand the needs of the user and provide a package that suits their method of working. I want a service provider to meet my needs, not for me to cope with their short-comings. The economics of the hotspot model (and the spectral limitations of 802.11b) preclude multiple operators installing hotspots in any one location. That means that the hotspot owners need to work together to provide global roaming between all of the available hotspots. I don't mind who my monthly or annual bill comes from, but I want it to come from a single supplier. I also want to have a good idea of how big that bill will be before it arrives. That's vitally important - for hotspot access to take off, the customer needs to have a reasonable idea of the ongoing cost.

The current situation is far from this. In the U.S. there are a number of hotspot companies that are rolling out locations, but most seem to have the aspiration of owning the entire country rather than co-operatively sharing the market. In Europe the network operators are beginning to embrace Wi-Fi as a revenue stream to complement GPRS and 3G. Although these are still national services, these are players that have extensive experience of the mutual benefits of roaming. It is interesting to see that we are also seeing competition from some fixed line PSTN suppliers, such as BT in the U.K..

An interesting middle approach is provided by the Internet Access providers GRIC and IPASS. These have always acted as independent clearing houses for modem and ISDN internet access for global travellers. Both have recently signed up Wi-Fi hotspots to extend their existing service. They continue to use their circuit switched model of charging on a per minute usage basis, which may prove to be unattractive for what is essentially an always-on connection. The biggest problem they face is that they address a largely technical audience and need to find a way of widening that exposure.

There is of course the free model of the coffee shop, which the industry loves to report. I still find it difficult to understand why a company that strives to increase its customer turnover wants to install a technology that would reduce the customer throughput. I recall an early experiment within Tokyo's MOS burger chain where every lunchtime the premises were filled with surfing geeks. The disgruntled customers queuing outside eventually gave up and lunched elsewhere. I never worked out who was the beneficiary.

There is an interesting variant of this approach that does seem to work, which is taking place in St Albans in the U.K. A wireless installer – Wialess has put Wi-Fi and laptops into a number of pubs. These are being targeted as free Internet sites for tourists and local business. By attracting tourists into the premises during slack times and extending the duration of visits at a location which normally has excess space, they are already finding that the provision of free Wi-Fi access is providing a 200% - 300% return on investment. There is presumably a limited number of sites in any area which can turn profit on this model.

The diversity of business models has resulted in a range of charging schemes which not only confuse, but vary enormously in cost. Many operators are aiming at a loyal user base by offering an annual subscription package, but the price points vary enormously. These packages offer unlimited access at any participating hotspot, although many limit the monthly data transfer to 500Mbytes. This is probably sufficient for general business use, although home and office usage is more typically 2–3 GBytes per month.

<1/> It is difficult to understand the difference in tariffs

Whilst the annual subscriptions are aimed at users who have regular travel habits, some tariffs are aimed at the less frequently peripatetic user. These are charged on a session basis and are frequently supplied as services to hotels and airport lounges, where travellers can sign up on an hourly or daily ad hoc basis. Given that we will probably not achieve true global roaming coverage there will always be a market for these, but they are inconvenient to keep track of and potentially expensive. They may end up in the same category as the airline seat phone – used either by those with money to burn or desperate to make a transaction. Some of the operators in this arena have made strategic moves to block competition by acquiring sole rights to popular locations. It remains to be seen whether this will be a profitable strategy or a stumbling block to development in the territories where these companies are operating.

<1/> Session based packages

A middle route, which could be attractive to the more casual user is a monthly subscription allied with a usage charge. These plans vary between charging on a per minute basis (suggesting a background of telephony billing mentality) or a data volume.

To compete with the session based model above these suppliers need to provide a good range of roaming locations to justify the monthly subscription. Without this the only advantage of the subscription is the fact that it guarantees you a consolidated, itemised bill at the end of the month. This is the area where both GRIC and IPASS have achieved success with modem access. However, they have been successful by selling their service to dial-up ISP's. They need to start looking at ways of bundling their Wi-Fi service to broadband suppliers, many of whom have a fixed line or cable infrastructure background, if they are going to emulate that success with wireless.

<1/> Subscription plus usage packages

It is instructive to compare these plans to see what they mean in terms of cost to a mobile worker. The following table takes published rates and computes the annual cost for a mobile traveller who access hotspots on ten days a month, with an average transfer of 5Mbytes on each of these ten days. Where a supplier has a variety of plans, the most cost effective is listed.

<1/> Cost to a mobile worker

The range in pricing is blatantly unsustainable. If hotspot usage is to take off in the millions it is generally agreed that the pricing needs to be between $25 and $50 per month. That is already being achieved by some suppliers. The irony is that those best able to market their services and grow a customer base are the ones with the most exorbitant charges.

Currently the suppliers most capable of handling the roaming issues are pricing themselves out of the market. These include the European Network Operators, who have the captive customer base and billing mechanisms. A new range of access technologies are appearing which will use the handset SIM to authenticate and charge wireless access, which will become even easier as the Bluetooth SIM access profile is implemented within handsets, allowing them to act as a personal authentication device for a Bluetooth enabled laptop.

If these operators grasp the opportunity and get the pricing right, they have marketing funds and the customer base to let them sign up large numbers of users in a very short time. As the user base is largely identical to the GPRS user base it allows them to develop a single data offering with Wi-Fi for hotspot locations and GPRS for seamless access in other areas. This ability to access and grow the user base gives them immense potential power. They could very rapidly achieve millions of users. If they do this it gives them the power to define the route of Wi-Fi development in place of the PC industry. Having installed 802.11b they will concentrate on maximising the return from this infrastructure and ensure that 802.11b remains the base-line standard for the next 5-7 years. But they must look at the competition and apply reality to their charging structure.

Voice over Internet; the key?

There is one further model which does not yet appear to have surfaced, which is to present Wi-Fi as a cost effective alternative to voice telephony. Mobile professionals who travel abroad are accustomed to phone bills of $30 - $100 per day, whether using mobiles or hotel phones. Using Wi-Fi as a broadband connection in conjunction with an internet or company VoIP gateway can reduce this to under $5. The payback for the user is almost immediate, and with a Bluetooth headset connecting to the PC, the process is extremely convenient.

This model is unlikely to appeal to hotels or mobile operators, who will both see it as a threat to their existing revenue streams. It does open up the possibility of the Internet VoIP services, such as Net2Phone and Go2Call to offer hotspot roaming as part of their telephony service. With the difference between hotel and international mobile calls and internet calls it could be a lucrative opportunity. Of all the competing hotspot aspirants, the mobile network operators are probably best placed to succeed. However, other distractions, such as their 3G commitments may prevent them from exploiting their advantage. All of the other players have problems of customer access and may have to resign themselves to either local custom, or piecemeal income form session based usage.

There are some indications that this may be challenged. AT&T Corp, IBM and Intel have recently teamed with two venture capital firms in an attempt to launch a new wireless access in a bid to create a nationwide Wi-Fi network. The enterprise – Cometa, has yet to announce its rates, but they are expected to be below the $50 per month subscription level. This will put pressure on the pricing models of the network operators, not just for Wi-Fi hotspots, but also for their GPRS tariffing models.

The effect on alternative Wireless data technologies

The relative markets for Wi-Fi and Bluetooth have already been covered and I see no reason why the boost for Wi-Fi will affect that analysis. The two technologies remain complementary. More importantly, mobile data usage begets mobile data usage. The more people who experience mobile data, the more there will be who demand it on a frequent and extensive geographic basis. Outside public access, which may play a part in overall coverage, Wi-Fi will only ever be implemented in locations where regular numbers of data users gather – the oft-cited airports, hotels and the like. And however hard the industry works it is unlikely that a single service provider will ever gain access to every commercial node. It is this "rest of the world" that will need Bluetooth and GPRS to maintain the data connection experience.

Whilst silicon vendors and analysts love to while away hours debating the relative merits of different technologies the mobile user remains unmoved by the different standards, being interested only in what works. The IT Manager rolling out mobile access will have a wider view of the cost effectiveness and Return on Investment, but once again, the specific technology is of relatively low import.

During the course of a day's travelling I will typically use a combination of Bluetooth, Wi-Fi and GPRS. It has always been one of TDK's tenets that they should work together seamlessly to deliver wireless data and that can be achieved today. The realisation that these technologies form a complementary set of tools for the mobile professional will help to develop better business models for data access, providing a service covering by combining all available technologies.

Conclusion

There is no doubt that Bluetooth is here to stay as a personal wireless technology, not least because of its support from the mobile prone manufacturers. I have never doubted the success of 802.11b as a wireless data access solution and Intel's promotion of the Banias will reinforce this position.

Over the next twelve months I envisage a stabilisation of the technologies that are used for wireless access, and once established these will remain the bedrock for a critical mass of users. That combination will be 802.11b, Bluetooth and GPRS. A few more bricks need to be added to make it fully acceptable to the market; there is a risk that WLAN adoption could be confined to the home if security remains an issue. That requires the industry to take a more mature approach in education - in particular promoting a better understanding of VPN's to give an adequate level of security across all of the radio connections.

Banias is a two-edged sword. It will create consternation amongst the wireless silicon vendors and probably hasten the demise of many in what is already an overcrowded marketplace. For the users, along with Microsoft XP's support for wireless networks it the immediate ability to present unwitting users with their first taste of wireless connectivity. That will change the aspirations of mobile users. It is up to the hotspot providers and Wider Area Network mobile operators to meet the demands of resulting business.

They have a limited time to do this. The penetration of community and free access points will grow apace. As more and more users experience free access the ability to charge will meet increasing resistance.

If individual providers continue to fight their own cause and no clear business leader appears it opens up an interesting alternative. The ultimate solution or Doomsday scenario, depending on your viewpoint could be for Microsoft to embrace the problem by adding hotspot roaming as a new MSN service. Then, whenever a wireless network is detected a subscribed user would automatically connect. It is a natural extension to Microsoft's vision of "connected everywhere." This lays down a challenge - unless the current hotspot providers can prove they have succeeded Microsoft may claim the mandate to make endemic wireless happen.

By Nick Hunn

References: Wi-Fi and Bluetooth - http://b-informed.co.uk/download/pdf/BluetoothWiFi.pdf Wireless LAN

“The Evolution of the Standard“ http://b-informed.co.uk/download/pdf/EvolutionLAN.pdf


Nick Hunn is chief technology officer at Ezurio, the Bluetooth specialist startup with the longest experience of any in the field