High-tech

Wi-Fi - bubble or bonanza?

By The Chilli analysts

The industry is in the midst of one of its worst downturns. Like the traveller who is stuck in a desert, it is easy to fall for the mirage of an oasis. Many technology vendors are looking to wireless local area networking (WLAN) to drive the next big demand for semiconductor devices. There are currently well over thirty semiconductor, intellectual property and component vendors participating in the WLAN market. Several European companies are involved including ARC, Infineon, Silaria, Synad and Systemonic. Investment in WLAN-focused start-ups Synad and Systemonic alone has exceeded $55M. The Chilli examines the WLAN market and argues that it is far from being the bonanza that we have been promised and is in fact a niche, awaiting a rapid expansion into a bubble.

1. Technical challenges
2. It's a technology, so what is the application?
3. Show us the money!
4. Feeding the chipmakers
5. The Intel factor
6. The Chilli perspective

1.Technical challenges

There are a number of technical challenges holding back the adoption of WLAN; firstly the fragmentation of standards, with 802.11b, followed by 802.11a and the emergence of 802.11g. Just which one should consumers pick? Is there a consensus from operators? Are users expected to pay extra for the luxury of a dual-band solution featuring a complex radio front-end?

Interoperability is an area familiar to most mobile network operators (MNOs). It could be argued that the relative success of 2G (GSM) was due to the MNOs, with their infrastructure provider, meeting the interoperability standards guaranteeing users from one network could talk to others with reasonable security. The slow deployment of Bluetooth technology is still hindered by the interoperability issues between wireless headsets, personal digital assistants (PDA) and handsets. WLAN has to go through the same growing pains. Various RF front-ends must work with various baseband devices - five RF front-ends and five baseband devices creates twenty-five combinations. To get all combinations of devices working together is not realistic, so operators will only approve certain combinations. Chipset vendors will not strive for full interoperability if it means that they could lose their hard won design slot to a potential competitor with bigger resources.

Wireless products generally facilitate mobility, but at the same time WLAN products have to achieve this within a power consumption budget. Power consumption must be reduced if WLAN chipsets are to proliferate in laptops and other mobile systems. Taking the concepts of interoperability and mobility further, the WLAN must facilitate roaming - what happens today when the user of a T-Mobile hotspot walks towards an area where the only WLAN coverage is from an Orange hotspot? Nothing - you just lose your T-Mobile connection. The inability to roam could be further compounded if different WLAN operators support different variants of the standard. A niche operator may offer a WLAN facility at Stockholm airport, but may not operate anywhere else. Do users carry several different variants of hardware and/or software?

Security has been a hot potato following the news that the wired equivalent privacy (WEP) protocol was insecure. This led to several companies banning the use of WLANs for fear that their data would be illegally accessed. A new encryption scheme, as per the temporal key integrity protocol (TKIP) of 802.11i, needs to be added to ensure that WLAN operators are not negligent in this important service criterion. This leads on to a concern of content providers in the form of rights management. In a network environment, content providers want to halt unauthorised use of their content, driving a requirement for digital content encryption, and a model for ensuring that content providers are paid for use of the content. This is an area that is still unclear for 3G services, let alone WLAN. Security issues for WLAN are still at the negotiating table, and any new standard will have to be implemented at a late stage, as it will be an afterthought, with a non-optimal trade-off between cost, power and security.

Quality of service (QoS) must be addressed, as the data streams must be handled without the user noticing degradation in data rate, video quality or audio fidelity. This is a challenge similar to that addressed by the protocols layered atop IEEE-1394 (Firewire). Another aspect of QoS is handling the co-channel interference between adjacent networks. Of course, while users are getting high-speed connections today, what happens when the number of users at a particular hotspot increases? Performance suffers. As noted by Tom Moore, business development vp at Silaria, "Alongside the acceptance of 802.11i, there needs to be more rapid movement on addressing the bandwidth and QoS issues through the ratification of new standards in these areas and the development and release of products".

What's the view of the service providers? According to John Lee, general manager of tetherless portfolio at BT Retail, the top three issues to be overcome to increase adoption are "Coverage, with sites in the right locations; security of the service; and service mobility, in terms of UK and international roaming".

2.It's a technology, so what is the application?

Wi-fi technology allows the transfer of data at various rates, depending on the standard used. That is just technology. What is the application that requires a wireless broadband LAN? For any communications medium, the applications are usually segmented into voice and data. Voice traffic is low-bandwidth, and is already conveniently served by existing handset technology. What about data? Data applications can vary from messaging, which is very low-bandwidth and well served by short messaging service (SMS), through to standard text documents, e-mails and web pages, right up to A/V streaming in the longer term.

Lets look at the environment in which the application is likely to be used. Wireless access would, not surprisingly, be great in a mobile environment. Leaving aside the stereotypical road warrior, what about the wider public? Would a typical consumer want to whip out their laptop to make a voice call, or would they have a burning need to transfer data to their laptop in the local Starbucks, whilst carrying two shopping bags and a laptop? Given that WLAN chipsets are still power hungry, do you also need to bring a power supply, or will your wireless flirtation last only several minutes? It is worth stressing that many people like to go shopping and have a coffee, but if they want to watch a movie, they would prefer to watch it at home on television or in a cinema.

What about a stationary environment, e.g. office or home networking? According to Kevin Mapplebeck, sales and marketing vp at Synad, "The initial market for WLAN as we are already seeing is connectivity within the business environment. Executives want to be able to connect to a network for e-mail and Internet connectivity as a minimum, and ideally, to their corporate network via a virtual private network, irrespective of location". This view is echoed and expanded by John Lee of BT Retail who stated, "In the next two years e-mail and synchronisation of data across wi-fi will be a principal driver for the increased bandwidth that's available. After that I think there will be customer sector applications like order systems and workforce management applications, and the need to have least cost routing over different access and roaming arrangements". Office networking is pretty well served in buildings that have desks with networking points. Passing road warriors are already used to hot-desking, and the low cost per node and reliability of Ethernet substantially undercuts WLAN. This would indicate that while there is a market in corporate connectivity, it is a niche. Depending on both the degree of accessibility and bandwidth required, some of these applications could potentially be served more cost-effectively by shorter-range wireless technologies such as Bluetooth.

As for home networking, a typical system would have one MPEG 2 programme stream, some digital audio and data being transferred around in real-time. This presents several challenges in terms of rights management and QoS. A bigger challenge for the home market is ensuring a low cost per node - $10 per node for WLAN, content encryption and QoS is excessive.

3. Show us the money!

Service providers will only invest in WLAN if consumers are prepared to pay for the service. What capabilities must a WLAN operator provide for paying subscribers? The commercial challenges to be resolved include infrastructure, customer service/support, billing and roaming.

A WLAN operator must have the expertise to select, obtain planning permission, commission and manage hotspot sites. It may be uneconomic to do this, as the potential number of WLAN users will always be lower than cellular voice users. This may mean that WLAN services are only available in densely populated urban centres. Even allowing for that, it will not be cheap, as premises owners are contributing to the WLAN bubble by charging operators high fees for attractive hotspot locations, such as airports, conference centres, hotels, major transport terminals and routes.

If an operator wants to charge, they must guarantee a level of service - that's hard to do with an unlicensed spectrum. What happens if an operator's hotspot is receiving interference from a nearby community WLAN? Who will provide the technical support if a traveller from South Korea cannot access the WLAN in a London Starbucks? If the problem is not resolved, the service provider loses a customer and Starbucks lose a sale and some brand respect.

The WLAN operator must be able to run a billing system, work with banks, perform credit checks, etc. The operator must also know how to bill for data, based on the type of content and the QoS - this is something that experienced MNOs are still struggling with.

4. Feeding the chipmakers

Many chipmakers have traditionally entered a market, looking only at the next step in the chain - their immediate customer - rather than examining the entire chain to see if it will actually generate any meaningful downstream revenues. If the operators don't invest and generate revenue, there will be nothing for others in the chain. An idea of the future for WLAN chipset makers can be gleaned by looking at what happened to those vendors who served the Ethernet market. During the 1980s, there were several Ethernet chipset vendors supplying to many adaptor card vendors. As the industry followed its natural path of integration, Ethernet functionality was subsumed into the motherboard and both card and chip vendors had to find another way to live.

To proliferate WLAN, the bill of material (BOM) cost needs to go from $30 to $5 and this will act as a spur to integrate RF and baseband functionality, thereby driving consolidation amongst the vendors, especially as few vendors have the capability to integrate on the required process technologies. Those companies with key intellectual property assets protected by patents will be of some value. Such a shakeout is reminiscent of the pains that Ethernet, GPS, USB and Bluetooth vendors went through.

For WLAN today there are few systems companies making WLAN add-ons. How will the chipset vendors feed themselves in the impending famine? Many won't. The shakeout has already started, with Micro Linear announcing that is pulling out of 802.11a. More announcements should be on the way.

5. The Intel factor

Intel Capital recently announced a $150M venture capital investment fund targeted at wireless broadband. This merits closer scrutiny. Intel Capital does not invest in seed stage companies, tending to be interested in co-investing in series A or B rounds. Although the fund is of a large size, it is not under pressure to invest all of it in a short period. Indeed, it will only invest in those companies who will support the WLAN infrastructure and help generate momentum for Intel's own products, such as StrongARM, suited to communications infrastructure, and x86, powering a large number of laptops. One could be forgiven for thinking that this announcement is a $150M tease to get other investors into WLAN, ultimately spurring demand for Intel products without Intel taking the majority of the risk. Intel may have trouble finding co-investors in the current market, and $150M does not go very far in building a network of hotspot infrastructure - just ask the existing MNOs.

6.The Chilli perspective

Back in 1989, Hutchison Whampoa introduced a cordless telephony service based on CT2 known as Rabbit. Subscribers to the service could make voice calls when they were within one hundred metres of a Rabbit transmitter. The service failed because people were forced to congregate around a limited number of transmitters to make calls, and mobile phone services that allowed roaming had just been launched and proved more popular. The Chilli opinion is that WLAN is the Rabbit of Internet data, given its weaknesses in standards fragmentation, interoperability, roaming and business model.

It is easy to forget that 802.11 is based on the use of unlicensed spectrum, so to make money from it is in fact violating the spirit of WLAN, especially with community groups setting up free networks and the practise of "warchalking", where chalk marks are made on the pavement close to a hotspot.

The expertise required to run a WLAN service is solely the province of MNOs, who have already made an investment in billing and service infrastructure. Their goal is to provide a more accessible service over a wider coverage in a consumer-friendly form-factor to a larger number of subscribers, promising a better return on their investment. If they haven't figured out how to bill for data services, they at least know how to bill for voice calls. MNOs should adopt WLAN services as a complementary offering to their cellular services, and offer it as a premium service to existing customers, funded from the MNOs marketing budget, rather than a separate profit centre.

WLANs will proliferate, but only as a narrow niche with just one or two vendor system solutions, and not the thirty odd currently in existence. As a niche the down stream revenues will be so small that it will be impossible to maintain a whole ecosystem of vendors including chipset vendors. Existing investors should take note of this in their next valuation.

This shift will have a tremendous impact, and many WLAN start-ups will fall by the wayside, although the vc community may be hoping that there is a potential exit of their existing investments through merger and acquisition activity. Given the patent stranglehold of established vendors such as Agere and Intersil, plus the activities of Atheros, Intel and Texas Instruments, what additional value do WLAN fabless chip start-ups have to offer in this market sector? If they are going to re-position themselves, which direction will they choose for themselves?

For further coverage of Wi-Fi, check out:

Due diligence on Wi-Fi fabless chip vendor Synad

Pwlan as a utility

Wi-Fi & Rabbit compared - by a former Rabbit employee

Comments on this story? Send an e-mail to Editor@TheChilli.com.

© Chilli Publishing Ltd 2003

15NOV2002