Copyright Ian Pearson, BT Futurologist

Click here for contact details,other articles and personal details

Voice and data integration

Ian Pearson, BT Futurologist, Nov 2004

People have been communicating for over a million years. Voice is one of the oldest mechanisms for human communications, along with body language, though who can say which came first? Long before we had the means to transmit voice further than we can shout, we invented the telegraph. This allowed data transmission, so were able to slowly transmit information in the form of text. A huge leap forward from there, and we could send audio, which of course we have used for voice. Nearly another hundred years later, we could send video too, so could finally send body language, albeit with poor quality. In a few decades from now, we may be able to send emotions, sensations and thoughts too. Although we think of audio, video and data in very different ways because of their significance to natural human communication, as far as networks go, they are all just alternating electrical or optical fields on some kind of wire. The same will be true of emotions, sensations and thoughts. They will be just another form of data too. It is purely an accident of technology history that we think of voice and data as something different as far as transmission is concerned.

Voice and data have been separated for so long because the old technology to transmit voice used analogue signals, while computer data used digital signals. Today, for home traffic, on old fashioned wired phones, voice is translated into digital signals as soon as it gets to the local exchange, and only converted back to analogue signals at the destination exchange. For mobile phones, for most cordless phones in the home or office, and ISDN phones, our voice is translated into digital information in the phone itself. Analogue voice is fast disappearing from the networks.

Of course, computer data is mostly transferred between machines to allow people to communicate, though machines do a lot of their own communications just to keep themselves working. Computers exchange ‘hello, are you still there?’ type messages with neurotic frequency. Computer data is thus a mixture of machine and human information, but is just a stream of bits. Now voice is pretty much the same, it is just a stream of bits too. There is no longer any real reason to separate them. As far as a network or computer is concerned, a bit is a bit. That is what voice-data integration is all about. It is the realisation that we can use exactly the same infrastructure to route every sort of communication, whether it comes from a human mouth, our fingers on a keyboard, by direct nervous system sensing, or from a machine.

Once we make this realisation, we can start saving costs and simplifying our communications management. Voice can be sent anywhere data can. We can have conversation via our computers as well as our phones. This integration has been going on in companies now for well over a decade. There have been a few generations of the technology, starting with Data-over-voice (Dove) in the early 90s, through to Voice over IP (VoIP) today. IP is Internet Protocol, and is the language used by computers to label packets so that they can be routed across the internet to any other computer. The internet is slowly spreading onto all kinds of networks that used to be quite separate. So now, people often use the internet to speak to each other via their computers.

However, even today, the phone networks are still largely separate from the internet. Most people and companies have two communications accounts, one for voice and one for internet access. Even users with ADSL or cable modems actually have separate channels on those wires for each of the services.

Today, they can even buy an internet phone that still uses the internet, but doesn’t need to be directly attached to their computer. It behaves pretty much like any other phone. But because the calls go via the intranet or internet, the charging mechanisms are different from when the calls go via the public phone networks or via a PBX. Internet subscriptions are often fixed price, with individual calls not being metered. Phone calls on the voice network are usually charged per call, per second, and even internal voice calls might appear on different parts of the balance sheet because they use different infrastructure. Sending calls via the internet can often therefore be cheaper. Additionally, for largely historical technology reasons, calls on the fixed voice network are transmitted as a fixed capacity channel, which is quite unnecessary now. Whereas they used to be allocated 64kbit/s for each call, all the time, in both directions, whether anyone is speaking or not, calls can actually be compressed down to as low a 4kbits/s with hardly any noticeable change in quality. (Mobile phone networks typically use about 8kbit/s to give perfectly adequate voice quality). While this saving couldn’t be realised on the old fashioned fixed network, it certainly can on the internet.

Because of these new opportunities for cost savings, big changes have been going on in the fixed network, which has been switching over to using IP for all calls. The fixed voice network is effectively becoming an intranet. The 21^st century BT network (21CN) will be fully IP.

VoIP treats voice pretty much like any other data on the internet, except that it labels the voice packets so that they get treated with high priority. Otherwise, they might have to wait around in queues at busy times, and users would sometimes hear gaps in the sound, which could make calls quite inaudible. It doesn’t matter much exactly when the various bits of an email arrive, but it certainly does matter when part of a word in a conversation arrives!

So voice data integration is progressing fast on the main networks. The low costs of using packetised data to carry voice, and of using advanced compression techniques are rapidly being implemented in the fixed networks, having already been rolled out in mobile nets. In company networks, voice can already be transmitted around intranets and LANs at reduced cost using VoIP technology. And when calls leave the company, they may also travel across the world on the internet. As the 21CN rolls out, it will become effectively just a high quality part of the internet. Callers won’t notice any difference in the quality of their calls, but the prices will continue to fall rapidly, as they have been over recent years.

Savings can therefore be made by going down this integration route. A company can have a single network and route all their traffic across it. With VoIP technology, voice calls take up a lot less capacity than with conventional technology, and can be mixed up with data traffic to give even greater savings. And with the new VoIP phones, the terminal equipment is still very familiar to users, behaving just like an ordinary phone.

In the future, our communications will increasingly use video too. This has failed to take off as a service so far, but much of the reason for this lies in poor bandwidth availability, high equipment costs and small screens with poor image quality and refresh rates. With full integration of all the media onto one network, lower bandwidth costs, integration of the service onto our desktop computers and cheap webcams, it is already much more attractive. People are starting to use it. Future technology will of course be better and cheaper. We expect that much of our communications will be in virtual environments, where we can see the people we are talking to in full 3D, as if they were there in the same office, with all the body language. We are even looking now at how we can capture signals from our nervous system to record sensations and replay or transmit those too. If you record what a handshake feels like, then you can shake hands on a deal with someone you are conferencing with. It will work with kisses and hugs too eventually, but that is not a matter for technology, but how well you get on with your colleagues.

1353 words