Features

Wireless LANs won't follow the path of hype

Even the most hardened sceptic would be forced to accept that wireless LAN is finally taking off. It seems that every analyst and manufacturer is convinced that it's the next big thing and the more different varieties that appear the better. Let's examine 802.11a, b, and g - and the role of hotspots, in determining their relative success.

Whilst I don't disagree with the benefits of wireless LAN I'd like to raise some questions, as there are forces afoot which could radically change the way it spreads into the market.

Unlike many of the pundits, I've been actively designing and using wireless LAN and Bluetooth for the last few years. Using the technology has the advantage of letting you discriminate between the hype and the real usefulness. The more I use it, the more I like it, but the less convinced I become of some of the predicted market trends. Make no mistake – Wireless LAN is here, and it's here to stay, but it could well follow a very different road to the one the PC industry analysts would have us believe.

So where are we? 802.11b has solved its major interoperability problems such that users can reasonably expect to plug devices in and get them to work. It's taken some time for the wireless LAN industry to learn the lesson that interoperability is important, but the formation of the WiFi certification program has proven the point and Microsoft's seal of approval by incorporating WLAN support within Windows XP has done the deed in turning the technology into a solution.

One of the reasons for that success is the fact that wireless does change the way you work. Most users of wireless networks will confirm that fact by refusing to give up the freedom it provides. They may still make wired connections within the office, but wireless becomes a way of life.

Although wireless is persuasive in the office, it's even more so in the home. Using wireless with a laptop has a marked effect on the way that users access the Internet. Always-on broadband technologies make it a resource that can be picked up and used at will, unlike the home PC, which entails a solitary trip to a machine that needs to be booted up, connected and then sat at for half an hour at a time to justify its existence. The biggest effect that wireless users report in the home is the change from infrequent, work-related access to a more topic-based, serendipitous approach. As well as enhancing corporate access it turns Internet usage into a far more social occupation - wireless allows the family to spend time sitting together in the same room again.

It's also beginning to be persuasive when you travel as more and more wireless hotspots become available. Throughout Europe it has surprised me how often I can find a network in airports and hotels. The problems of roaming and billing still need to be addressed, but the infrastructure is starting to appear.

That doesn't mean that wireless LAN has solved all of its problems. The security shortcomings (which I personally believe highlight the shortcomings of many pundits rather than the technology) have been widely reported. Why allegedly informed IT managers see nothing wrong with entrusting unencrypted data to any form of physical link is difficult to comprehend. This has nothing to do with wireless. How many people happily connect their laptop to a phone or internet point in a hotel room? It requires almost no technical expertise for an hotel employee to tap into that call. By now it should be basic common sense that any data transfer outside a secure company network requires the use of a VPN, whether it's via PSTN, ISDN, GSM, but still we see the belief that security is down to the final physical link. The presence of a wire does not make communication safe.

This whole attitude is reminiscent of what happened when companies discovered that documents could be stolen from dustbins – the solution is not to put the confidential documents in a black plastic bag and hope the dustbin lid is secure – the answer is to shred the documents first. So let's not blame wireless standards for security problems.

With a bit of luck common sense and education will remove the security paranoia. It's debatable whether the short term improvements within WEP and 802.11x will prove to be any more secure, so the industry really does need to take the VPN route to put this issue to bed. What is more concerning is that the wireless LAN industry is doing its best to shoot itself in the foot by launching a new raft of standards just at the time when users are beginning to accept the current one. We're also seeing a divergence in the forces driving wireless acceptance. Up to this point it has been firmly driven by corporate networking and the PC industry. We're about to enter a new phase of development where hotspot deployment could become the primary driver.

Alphabet soup

Wireless LAN may be a good thing, but like many good things, there are too many different varieties appearing. The first new entrant is 802.11a. 802.11a started life before 802.11b, but has taken longer to arrive because it involves a more fundamental redesign of the radio. The rationale behind 802.11a was to provide a higher data throughput, giving raw data speeds of up to 54Mbps. This correlates to a real data speed similar to that experienced by most desktop users.

Anyone contemplating wireless networks should be aware that the raw data rates you see in the literature are largely unrelated to what you'll get in practice. By the time the overheads of control signals and error correction have been taken into account the maximum data rate is around half of the headline figure, and that bandwidth will typically be shared between half a dozen users. Nevertheless a useable 20Mbps is better than most 10Mbps wired networks, which makes 802.11a a very attractive option.

The other aim of 802.11a was to increase the efficiency with which the radio spectrum was used, allowing more wireless networks to be co-located. A frequently ignored aspect of 802.11b is that it only allows three co-located networks, which can be a severe restriction in small offices or flats. 802.11a increases this to 12 in the U.S. and 19 in Europe.

To achieve these two goals, 802.11a has been implemented at a higher frequency of 5.1GHz, compared to the 2.4GHz of 802.11b. In order to squeeze more data out of the band, 802.11a has implemented a new, more complex coding scheme – OFDM (Orthogonal Frequency Division Multiplexing). This more complex coding, plus the inherent difficulty of designing silicon chips at twice the frequency has allowed 802.11b to come to market first, despite being second off the starting block. It has delivered on its promise and has gained acceptance as the first widespread wireless LAN standard.

Incidentally, 802.11a has an added advantage over 802.11b by using the higher frequency. There is considerably less interference within the 5.1GHz band, because it is little used. In comparison 802.11b shares the 2.4GHz band with Bluetooth, U.S. cordless handsets, microwave ovens and a host of consumer devices such as baby monitors and video senders.

Although everyone knew that the 802.11a development process to take a considerable time, most of the industry had expected 802.11b to arrive a few years earlier, giving it time to settle in and be accepted before 802.11a burst onto the market. This timetable ignored two things: the inevitable delays in getting interoperability sorted out within 802.11b, and the commercial climate in the late 90's where every silicon designer with an eye to future riches went and acquired venture capital to design 5GHz chips. The result is that instead of a dignified progression of technology, 802.11a has burst onto the scene at the point that 802.11b is only just starting to see itself established.

For the user it means confusion. Do they deploy 802.11b or do they wait a few more months for a faster 802.11a? The problem with wireless LAN is that it requires a moderately high investment in infrastructure, so making the wrong decision can be expensive. There's been a rash of announcements offering “upgradeable access points†but a distinct lack of explanation of how that upgrade path will happen and how much it will cost. An added problem is that today there is only a single silicon vendor for 802.11a – in the short term that should ensure interoperability, but in the longer term there is bound to be a period of instability as new entrants begin to ship their products. And already we're seeing proprietary extensions with companies offering go-faster versions with double the speed. But these faster speeds are only achievable if you buy the one brand, which makes you wonder what those manufacturers think the word “standard†really means.

Who needs 802.11a? In the short term very few. It provides the bandwidth for streaming video applications, but at the moment it doesn't contain any provision for Quality of Service. That will come later when the contents of 802.11e are implemented. In a congested office 802.11a certainly provides a significant performance advantage, but if you are travelling and likely to access a public LAN then it will be incompatible. My belief is that 802.11b will be the standard that is deployed for global mobile access and that 802.11a will remain a novelty until such time as stable, interoperable, multimode 802.11a+b products appear on the market. Even then almost all wireless access will still be via 802.11b.

One area which seems to be ignored for 802.11a is wireless LAN for desktops. As a USB 2.0 connection it has the potential to ease standard desktop connectivity, and could become immensely popular and cost effective once the retail price drops below the $50 mark. For this reason we predict a significant increase in .11a sales starting in 2005.

Overall I am less bullish than many analysts about the introduction of 802.11a. I firmly believe that 802.11a will be successful over time, but in the short term it will remain a technical curiosity, taken up mainly by those who must have the latest and fastest.

In fact the early introduction of 802.11a may even help to promote 802.11b. Bizarre though that may sound, it is based on the observation that the silicon vendors involved in driving 802.11a are currently different from the established vendors of 802.11b. If, as seems likely, the demand for 802.11a is low, the companies in this market will start to reduce price. Because they have no exposure to 802.11b they will have no concerns over cannibalising existing revenue and are likely to opt for aggressive price points. 802.11b suppliers will be forced to react by reducing the cost of their solution, as the PC marketplace insists that slower technologies must be cheaper. It does not need much of this pressure to cause the price of a WiFi adaptor to fall below $50, and at that level it could become as popular an accessory as the modem was. Although this is good news for the user and the Hot-Spot providers, it suggests that the chance of manufacturers making money from wireless LAN hardware in the PC market is very slim indeed.

Partly because of this, 802.11a seems to be attracting more real interest amongst consumer electronics companies for video transmission, where it continues to battle with HiperLAN/2. There's still life in HiperLAN/2 and if 802.11a fails to gain enough market share in its early stages we could yet see a decent battle.

The role of Hot-Spots

Almost all analysis of the WLAN market is PC Enterprise and SoHo Centric. This report is not attempting to look at the wider consumer market, but would like to address the effect that wireless hot spots are increasingly likely to have on the continuing usage of 802.11b. Because these are likely to be implemented by the telephone operators, who see them as an extension of their existing telephony revenue streams, their effect has been ignored in many of the WLAN surveys. That is a dangerous omission, as ubiquitous 802.11b access for the mobile worker could be a more significant driver than the office.

The last twelve months have seen a renewal of interest in wireless LAN to provide hot spots of connectivity. The initial business models of free wireless access in every Starbucks has been somewhat discredited with the business failures in this area. In its place we are seeing increasing confidence among the telecommunications suppliers to start providing wireless access for their customers, both from fixed line operators and mobile phone networks. These are all working on a business model where wireless data access becomes a supplement to the monthly bill of existing customers.

I predict that many of these business models will succeed. These are communications players who already understand the mechanisms of roaming. Roaming is vitally important because users will not want to pay a separate charge for every location, which is the current model. In addition there are not enough users to support multiple access points at any one location, and the limited three channels of 802.11b effectively prohibit multiple ownership in any case. Many of these operators are desperate to pursue easy methods to increase ARPU (Average revenue per User), as they see 3G deployment recede into the middle distance and perceive WLAN as an ideal candidate to achieve this goal.

Telecoms operators do not like changing standards. They have had their fingers burnt too often with new technologies, having been bitten by WAP and GPRS. When they enter this market they will want to do so with proven, stable standards which will have a five to eight year shelf life. It will be almost impossible to convince them to adopt a new and essentially unproven WLAN standard. That means that the majority will deploy 802.11b and not 802.11a or 802.11g. They are already doing it and will probably still be doing it in five years time. The growing base of hot spots will re-emphasise the need for 802.11b amongst users and extend the life of the technology far past the projections of the 802.11a advocates.

An early indication of this trend is Nokia's recent announcement of their infrastructure access point – the A036. The design of this unit has been optimised to make it software upgradeable, rather than hardware upgradeable, illustrating how well they understand the mentality of the carrier. What is seen as important for an infrastructure product is the ability to cope with future concerns such as 802.11x, 802.11i/TRKIP and Extensible Authentication Protocols (EAP), not with new radio standards. I firmly agree with Nokia's approach that these security and payment led issues will prove to be far more important than connection speed.

As these hot spots become more numerous, they will further reinforce the wireless access model, persuading many laptop users to experiment with wireless. For many, wireless connectivity outside the corporate will be their first experience of 802.11. By 2006 we predict that almost half the users of 802.11b will be connecting to hot spots whilst they travel.

Global Considerations

Much of the focus and analysis for 802.11b has come from the U.S., where the standard was rolled out successfully. The euphoria surrounding this success has masked the fact that the 802.11 stable of standards are not truly global, particularly when the 5.1GHz band is considered.

Between Europe, Japan and the U.S. there are some subtle and not so subtle differences in radio regulations, that mean that no 802.11a product is usable throughout the world. It has already been noted that the channel allocations differ in each of these territories. It is not just that the available number of channels is different - the problem is compounded by the fact that the radios need to transmit at different frequencies.

Within Europe there is a further complication because radio regulations require better etiquette in using the band. Because other systems also use the same spectrum, European regulations require all wireless LAN systems operating at 5.1GHz to implement a form of Dynamic Frequency Selection (DFS), where they look to see if someone else if using the band before starting transmission. Once clear spectrum is found, they must employ Transmit Power Control (TPC) to ensure that the connection is run at the lowest possible level. The approach is similar to the human norm of listening before you speak. Conversely the FCC allows units in the U.S. to shout without listening first.

These issues of TPC and DFS are being addressed by the 802.11h standard, which adds them on top of the existing 802.11a. This should be ratified within 2002, allowing units to be designed which could be global. However, anyone buying 802.11a before this point may end up having to replace their hardware.

This last issue poses a problem for WECA – the organisation trying to promote interoperability. They have successfully launched the WiFi logo for 802.11b products and are currently grappling with the meaning of their WiFi5 logo for 802.11a. If they allow this to be used for current U.S. products, these will not work in Europe and the user perception of WiFi5 as a token of interoperability will be destroyed. The sensible approach would be to wait until 802.11h is ratified and implemented, along with a harmonisation of frequency bands throughout the world. In this way WiFi5 would be a global statement of interoperability. However, WECA are under pressure from the U.S. industry to relax this global approach to pursue a purely local approval – a move which further threatens to confuse users and delay the adoption of commitment to 5.1GHz.

802.11g – The standard we don't need

As we've said before, in order to take off, wireless LAN needs to be seen as simple. Users don't want confusion or interoperability problems. Unfortunately one part of the standards body has forgotten this and is threatening to wreck user confidence with the “interim†standard of 802.11g.

802.11g was a good idea at the time, when the road to .11a silicon looked long and arduous. However, infighting between its proponents has delayed .11g to the point that it is now irrelevant and deserves a mercy killing.

The work on 802.11g was started to see if the data rates of 802.11b could be pushed up to 22 or 54 Mbps. During the course of its development it turned into a commercial battle between two major silicon vendors, each attempting to promote their own specific technology. After twelve months of delay the 802.11g committee appears to have realised that some form of decision was necessary if only to preserve their credibility and continued existence, so issued a draft proposal based on a compromise with optional coding schemes. That draft compromise still contains a lot of holes and “what if'sâ€. Even if these are addressed the standard is still a year away from completion and by the time the first products appear 802.11a should be established as the high speed WLAN option.

There is a more major concern for 802.11g. By the time it arrives there will also be over 100 million Bluetooth devices around, all using the same chunk of spectrum as 802.11g. They'll provide a level of interference which ensures that 802.11g will fail to offer its potential increase in speed – it will just keep on stepping down to 802.11b data rates. There will also be around 20 million 802.11b devices in service and the backward compatibility modes of 802.11g ensure that if they're detected 802.11g once again slows back to 802.11b rates. In other words the only place where 802.11g products will attain their promised higher speeds will be inside a screened box.

Which means that there's not a lot of point in going any further with 802.11g. Unfortunately there are a number of supporters who still believe it's the cleverest thing since sliced bread. From a selfish marketing point it may be – I'm sure there are a few companies who will increase their market share if it ships, but for the industry as a whole it could be disastrous. At a point where the end-user should be feeling comfortable about wireless LAN it throws in yet another variant to confuse and delay.

In common with 802.11a, 802.11g is not legal in Europe (in fact it's still debatable whether it's approvable within the U.S.), so manufacturers will need to lobby for radio regulations to be relaxed. That raises a worrying scenario – both 11a and 11g will be petitioning for a change in the rules, both for the same user benefit – higher rate wireless networking. There's a good chance that the regulators will decide that there's no sense in relaxing two different bands and ask which one is the more important. If that happens we could see two wireless LAN industry groups fighting each other for the moral high ground. The decisions will get delayed, the momentum behind wireless LAN roll-outs will falter and the only eventual winner might turn out to be HiperLAN/2.

That's probably an extreme scenario, but the reality is that once again the industry's enthusiasm to roll out new technology has resulted in the user being totally forgotten. The security issues over 802.11b show how insecure the corporate market still is regarding wireless. What the industry needs most is stability – a useable standard that can be deployed and which works.

Conclusion

Wireless LAN is here, but it's still in its infancy. If it is going to be successful the industry needs to start thinking about customer needs rather than technology for the sake of technology. Networking is a long term investment, not a fashionable toy to be changed on an annual basis. It is useful to remember that corporate culture has only deployed two different network standards to the desktop in the last twenty years – the offer of three different wireless standards in eighteen months is not going to be a welcome message.

What users want is to be able to connect to a wireless access point in as many places as possible. To achieve this the industry would do well to concentrate on making 802.11b the standard wireless network for at least the next five years. The increasing interest in hot-spots is bringing the influence of the telecoms industry into the equation with their emphasis is on stability of infrastructure. This is likely to have a radical restraining effect on the more recent flavours of the wireless standards. At the most extreme level, the telecoms operators could even wrest mass deployment of wireless LAN technology away from the PC industry.

Nick Hunn is Managing Director of TDK Systems Europe Ltd

Reference: Bluetooth and WiFi - the market status. Nick Hunn - Mar 2002