As already stated, the eventual saturation of the GSM networks is a guarantee that the 3G networks are going to be built, irrespective of the way that the 3G licence auctions have slowed investment. There is an argument however, that the longer it takes for the 3G service to be established, the more the potential profit base will be eaten away by alternative technologies. Enter, stage left, the so-called unlicensed wireless technologies.

As the name suggests, unlicensed wireless technologies are short range transmission frequencies that do not roam and do not have to be licensed from the government - and therefore are immune to the kind of heavy indirect taxing we saw with the 3G license auction.

Short range unlicensed wireless works on the premise that you shouldn't need a long range, expensive 'Martini-style' service that can connect you 'anytime any place anywhere' for effective mobile computing. The Holy Grail here is a cheap, mobile connection with fat bandwidth that, when used in conjunction with a base-station, can offer much of the benefits and mobility transported over fat, cheap fibre pipes.

The user can then access whatever they want from the internet using all existing mobile-enabled applications.

802.11b

802.11b is a short range wireless technology that has been seasoned over five years in the commercial marketplace and can deliver a healthy 2Mbps (although it promises 11Mbps, actual experience of this will be much lower, depending on access to the base-station and other factors). The silicon is now cheap enough to be integrated into notebooks as an attractive value-add - a very useful way to share files over a peer-to-peer connection, especially if you're part of a roaming team of engineers. 802.11b has also made inroads into the warehouse, integrated into bar code scanners and other handheld devices by companies like Symbol and Psion Teklogix.

In the US 802.11b style LAN services have become increasingly common in hotels and airports. Earlier this year Starbucks, the MacDonalds of the coffee franchise, announced they would be adding 802.11b services to their outlets. Bearing in mind the ubiquity of Starbucks outlets and the crippling incompatibility between the patchwork of licensed mobile frequencies across the US, this is an attractive opportunity for businesses with widely roaming mobile sales-forces. No matter where sales are visiting, it is almost guaranteed that a Starbucks will be nearby. All a user of an 802.11b-enabled notebook needs to do is walk in, buy a coffee, boot their notebook and log-in to the Starbucks LAN, and they are then able to reach whatever applications their company has transcoded for browser access.

Of course, the service is more suited to the US and its culture of free local call rates than it is to the UK, where there are a number of legal wrinkles that need to be ironed out. 802.11b falls within the ISM industrial, scientific and medical band of 2.4GHz. This is a blanket licence in the UK that is unregulated and free to use providing it is not exploited for commercial purposes. In layman's terms, this means that a company must own both ends of the wireless connection - selling an open connection and related service is technically illegal.

There are workarounds, which usually involve making the connection 'private' by bundling it with another function, such as a frequent flyer or hotel services loyalty card. The UK Radio Communications Agency apparently has stated that reform is on its way, but this will is going to take 12 months to get going at the earliest.

Setting up an 802.11b wireless LAN is far, far less complex and expensive than setting up a standard ethernet LAN. All it involves is purchasing a base-station, a cheap hub and as little as an ADSL connection. Correctly configured, the base-station can provide multiple and dynamic connections to 802.11b enabled devices across shared ADSL bandwidth.

On the other hand vendor interoperability has been poor in this field and 802.11b solutions, although ubiquitous and cheap, can be difficult to support and manage, and the difficulties are set to increase as more devices start using the 2.4GHz frequency.

802.11b delivers the 11Mbps data rate by taking the 83MHz channel and splitting it into three 22MHz wide channels, a process called Direct Spread Spectrum Sequencing (DSSS). Unfortunately, this may increase the data rate, but it makes 802.11b more prone to interference from the increasingly crowded 2.4GHz space - cheap microwaves being the most frequent domestic offenders.

HomeRF has been around even longer that 802.11b and uses the same frequency. HomeRF transmits a lower data rate over 79 1MHz channels, but has more room to hop frequencies if it hits any interference (Frequency Hopping Spread Spectrum). Additionally two of these channels are dedicated to isochrynous channels, or voice transmission. The silicon and tools for Home RF solutions are widely available, having been used in DECT cordless phones for years. This gave HomeRF the advantage over 802.11b, which was three times as expensive to produce, until the last six months, when it seems the silicon price and availability of the latter has dropped to a competitive level.

Bluetooth

We've already mentioned Bluetooth in the context of handsets, but Bluetooth is often cited as an unlicensed network solution. Two years ago the Bluetooth network stack was third verse, Chapter 1 of the Bluetooth Evangelical handbook. Developers have, however, been hard-pressed to produce a working version since the early days of the blue sky theorising. RedM, the spin-off from Madge Networks, has been more successful than most, and won the Innovation award at the Bluetooth Conference in Geneva last year. The mythical Bluetooth network drop, which can theoretically support a 432.6 kb/s symmetric link, or a 721 kb/s asymmetric link, over a 10 metre range.

With the necessary investment in technology, this solution, in the form of Bluetooth enabled phones, could potentially offer cheap international voice calls courtesy of VoIP, and, in the case of Bluetooth enabled notebooks, very cheap access to the corporate LAN.

There are obvious and serious questions as to how either service will make money, and who will get their hands on it. Some readers may remember Hutchinson Wampoa's precedent of the Rabbit network, which was hugely profitable in Singapore but spectacularly failed to deliver a similar service to the UK in the 1980s.

The compelling interest in Bluetooth in the unlicensed space seems to centre around issues of interference. The common opinion from the pundits seems to be that Bluetooth, with its rapid and aggressive frequency-hopping rate, will crush 802.11b signals.

Various vendors and semiconductor designers propose solutions to this interference problem - whether the interference comes from outside or whether both solutions are present on a single device. These solutions tend to either implement a technique called 'fast-driver switching', or try and be completely silicon based. Either way, it is unlikely that the first generations of Bluetooth and the current generations of 802.11b will benefit from them.

Other commentators believe that, for these first generation devices at least, it is the size and strength of the transmitter, and in turn the size of the battery, that will decide the outcome. The 'might is right' theory says that a notebook with 802.11b will have the stamina to take down a Bluetooth on a mobile, while a PA system will drown out the two. The interference issue is likely to be refined away in subsequent generations as a result of diligent lab work from all the major vendors.

802.11a/HiperLAN

Although some analysts have estimated the market to be worth as much as billion by 2005, it doesn't really look like the 2.4GHz band is where the big money is in the long run. This belongs to HiperLAN2 and 802.11a technologies in the 5GHz band, which is the real sweetspot in silicon development just now.

802.11a and HiperLAN2 are two standards, the former developed in America and the latter developed in Europe. Essentially they use the same silicon, but a different configuration in the MAC layer. Both the solutions transmit on the 5GHz band and offer potential thoughput of up to 54Mb/.

2.4Ghz is destined to become overcrowded with Bluetooth, DECT and 802.11b devices, which are incapable of efficiently negotiating their way through the scrum without the expense of independently developed solutions.

The 5GHz technologies, on the other hand, use a technique called Orthogonal Frequency Division Multiplexing (OFDS) to maintain a very high bandwidth in a very noisy environment. 5GHz silicon is expensive, but the price is falling as Silicon Germanium and Silicon Arsenide processes are integrated into the fabrication plants.

Although 2.4GHz solutions serve their purpose for introducing the concept of short range wireless to industry and public the serious investors are likely to be looking to 5Ghz as the technology which can deliver their services, and governments have a much better chance of modifying archaic legislation before they start filling the airwaves.