So why is there a push for voice over IP over wireless (VoIPoW)? And why is it happening in the wireless world where good voice quality is harder to achieve and the air interface protocol has long built in delays? Why is this not happening in fixed line business where the big telcos also operate? Every home PC can easily be used as a voice over IP (VoIP) terminal, but big telcos and big infrastructure vendors are very quiet about it.

Well, when the boat is moving more or less steadily, nobody wants to rock the boat. Fixed line VoIP enables everyone to make multiple overseas and long distance calls for one local call price, only in the wireless world can operators still control call prices per application used. Big telcos are not so afraid of VoIP over 3G, but a world wide fixed line toll-quality VoIP services would be catastrophic to revenues.

The wireless industry is about to have the biggest change ever. When a big change is in a horizon, new players want to shape the future to their advantage. Companies with strong IP background and not so strong 2G backgrounds want to increase their wireless market share and are strongly pushing new alternative ideas. Once 3G operators are cemented into a long-term supply contract with equipment pricing schedules and vendor financing chains, the course is set for the chosen vendor's technology. Make no mistake, these decisions will be career decisions for the directors and will be a sink or swim situation for a wireless operator. Even excellent services and applications will not save the operator if the information pipe can't handle it. Many countries now have a wireless number portability system, meaning that if a customer is not happy with the price or the quality of service, he or she can switch quickly to another carrier.

A “Wireless VoIP only” road map sounds like the GSM half-rate story. In the early 1990s all mobile infrastructure vendors had GSM Half Rate coding in their road maps. The plan was to implement the half-rate coding to add network capacity through better spectrum efficiency with a comparable voice quality. Most of the network operators were demanding this in 1994-96, but when half-rate was ready to implement, nobody wanted to be the first one to have it. The main reason given by operators was that they did not want to implement something that would reduce the voice quality, when Enhanced Full Rate (EFR) voice coding was coming around the same time. The first mass EFR implementation happened in 1997 and after that half-rate was forgotten. GSM voice quality and clarity become a major advertising topic in 1997.

So what is the problem with wireless VoIP? In short, it is still under development with a goal to match circuit switched voice quality using similar spectral efficiency. All mobile networks are optimized to give the best possible voice quality through the bottleneck, the air interface, where every bit of extra capacity counts on the bottom line. VoIP will either add extra headers into the air interface, or packing and unpacking has to be done in the core area using an additional network element. If voice packets are passed through the air interface, even with the compressed header technology, at least 10 per cent of voice traffic contains voice packet headers. If packing and unpacking is done in the core network area, it will add some time delay without improving the voice quality, thus some flexibility of the IP service on offer will disappear. Packet delays and handling of lost packets will also problematic. Coding delay (if packet is generated in a network side), network delay, jitter buffer delay and queuing delay can add up, especially when calling overseas. At the moment, sending a packet half way round the world and back typically takes 500ms. Even if that time slightly decrease in the future, it still does not match the existing voice lines. When you add 3G voice coding to that, with two levels of interleaving at both ends, it will feel like an eternity to a mobile user.

Competition will also be there in the VoIP world. Probably the greatest threat will come from wireless LANs. A very cheap localized wireless LAN (802.11b) can offer a cheap VoIP alternative with up to 11Mb speed and 100m cell range with Internet and voice access points. It is already widely used in educational institutions and the corporate world as a great cost saving application and operates on a free unlicensed frequency band. Global ISPs, multinational companies and even public groups and initiatives can build very cost efficient networks in metropolitan areas. All you need is a couple of antennas near Internet cafes and other shops, and a downloadable cheap access point map. Wireless Metropolitan Area Network (802.16) for “the last mile” is promising even higher speeds and will be competing for the same wireless high speed corporate market data revenues as 3G networks. Are 3G operators willing to compete with them in the wireless VoIP market after paying for an expensive license, financing and infrastructure?

But is this entire debate just storm in a tea-cup? Has anyone asked an average mobile user if they want voice over IP or circuit switched? The answer would probably be that he or she doesn't care as long as it is has a better service quality and doesn't cost much more than the current system. People care about mobiles, applications, quality and cost, not the underlying technology. Given the current 2G networks voice quality problems - high call drops, interference spots, poor indoor coverage, shocking call performance in high rise urban building - it is hard to imagine selling the current level of voice quality as a premium service. High-end users will demand it as a minimum. Large corporate clients will start to demand quality of service performance guarantees of voice and data in all traffic conditions. This will be a big task in WCDMA systems capable of delivering fluctuating high speed data along with a constant low speed voice data, especially near the cell edges where “cell breathing” happens. Technology oriented corporate clients will make sure that the tea-cup is getting bigger and the storm demanding better quality service will last a long while.

The building of 3G networks has already started in competitive and financially tough environments where mistakes are not tolerated. In these new networks wireless VoIP will be provided as a part of the 3G multimedia applications. Some Eastern European countries have already successfully experimented VoIP with GSM. With cheaper call rates, wireless VoIP will definitely have its place in the future, when lower voice quality is acceptable. Initially all voice connections in 3G networks will be circuit switched. 3G VoIP experiments will follow soon, but do not expect 3G operators to go VoIP only, any time soon. Meanwhile 3G vendors can show their product development by making their international 3G network sales calls over wireless IP. While we are waiting for that, it still looks like one of the 3G killer applications, the traditional voice traffic, will remain in circuit switched side to guarantee the quality.

- Petri Possi is an independent wireless network consultant. He has over ten years experience in a wireless industry working with network planning companies, infrastructure vendors and mobile operators. You can send him feedback through  info@umtsworld.com