Brave new future

What will IT have achieved, five years from now? Computer Weekly approached some of the more creative corporate research and...

What will IT have achieved, five years from now? Computer Weekly approached some of the more creative corporate research and development establishments and asked them to preview their most interesting projects. Andy Favell highlights a few that caught our eye.

Touch and feel: BT Exact

Scientists at BT Labs are working with "haptic" interfaces that allow people to touch and feel something remotely.

Before you get excited at the prospect of upgrading long-distance relationships, that is not what BT Exact had in mind.

Haptic devices may look like joysticks, pin boards used for modelling, mechanical gloves or robots to the untrained eye, but they give the sensation of touch as a camera or sensor unit is moved over a remote object (there is a gallery of devices at

One of many applications of this technology could be to help partially sighted people to "see" something online, perhaps by reading Braille. It could also be used in remote hospital operations, where the sense of touch is important, or to allow medical students to experience conducting surgery as if at first hand. If prices come down considerably future shoppers may be able to feel items in the online shop.

BT's development work is focused on making haptic interfaces work over the net. "There are issues in implementing this over a packet network," says Steve Wright, head of long-term research at BT Exact. "It requires just the right amount of information to be delivered to the device to prevent instability." If too much data arrived at once, the effect would be magnified and your hand would be pushed away.

Data tagging: HP Labs

There is a distrust of applications at HP Labs and a belief that information should be secured at the operating system level.

Scientists at HP Labs claim to have invented an easy method for applying an irremovable data tag that remains with the document or any part of it through every change or copy, no matter what the application.

Checks at the e-mail, print or web server, for example, will spot if tagged data, whether text, pictures or video, is being e-mailed, posted to the internet or printed off, against the rules.

"While a lot of work has been done in this area, in the past the results have been systems that are overly restrictive and ultimately unworkable," says Chris Dalton, research engineer at HP Labs.

He argues that HP's approach does not have these drawbacks and, over the next couple of years, tests will show the business scenarios where the technology will work best. This may allow confidential documents to be shared between organisations.

Natural language search: Xerox Parc

It is not the coolest name, but the Knowledge Extraction from Document Collections (KXDC) team at the Palo Alto Research Center has a bold, long-term vision "to build computers that can acquire and reason with information that is expressed in natural language, and can communicate in natural language on a par with human peers".

Reinhard Stolle, research scientist at Parc, says, "This is clearly an extremely ambitious long-term goal, but we have many ideas about how to create valuable intermediate milestones along the way, and our results are very encouraging."

In the future, users will converse with the video recorder, saying, for example, "Record Frasier tonight. VCR: there are two episodes tonight. The one after the news."

In business, natural language will help companies such as insurance, banking, legal and pharmaceutical firms to search through huge databases of information and organise them by identifying redundant documents. Stolle believes that it could improve customer relationship management and content management applications.

An early incarnation of this technology is used on a database of technical documents relied on and constantly added to by copier and printer service technicians at Xerox.

The key is to teach the computer to recognise "aboutness", so the system can tell the difference between documents that are, say, pro- or anti-war in Iraq. A user could request a news site to send a document that confirms a merger when it is published, without having it send speculative articles.

Related projects at Parc include using natural language to improve computer translation and enhancing techniques for scanning huge libraries of documents and turning them into digital documents.

The KXDC team expects elements of its work to be ready during this decade.

Ad hoc networks: BT, Intel, Cisco, Xerox and others

The most common application for the ad hoc network or a network-in-motion is on the battlefield or between emergency vehicles. Imagine a battalion of moving tanks using each other as network hubs, perhaps using a wireless Lan to communicate, not needing to rely on a fixed network or mobile phone network at all.

There are simple pilots of mobile networks around the world, including one where taxis are sensing and reporting back weather conditions in Japan, but none of them are truly network independent, says Alain Fidocco, director of technology marketing at Cisco. The first trials of true ad hoc networks, he suggests, will be in the next 12 to 24 months.

Cisco has developed a mobile router about 5cm square, but miniaturisation is not the hard part. The serious research is going into reinventing the network routing mechanism. The problem is that fixed networks are anarchic, there is little sense in how data gets from A to B. That cannot happen in a network with a continually changing geography, otherwise looping will be inevitable.

At this point expect the BT experts to start talking about the cellular make-up of fruit flies. Suffice to say BT has picked up an algorithm from the fly that could help these self-organising networks to organise themselves. QinetiQ is examining how BT's algorithm could help in the battlefield scenario.

IP caller ID: Bell Labs (Lucent)

Europeans do not get as hung up about knowing who is calling on the telephone before they answer as people in the US do, but we would all certainly like to know what data is heading for our network.

The problem is that it is easy to forge IP addresses and, at an application level, e-mail addresses, to mask the true origins of a message. Lucent Technologies' Bell Labs believes it is working on the answer: IP caller ID.

"I should include information in the IP address of a message to show where the data is coming from, but there is nothing in the network that forces you to abide by that regulation," says Wim Sweldens, Bell Labs' vice-president of computing sciences research.

Carriers could implement ingress filters, which check that IP addresses match origins, but they have been dragging their feet. Suspecting that it could be a long time, if ever, before all networks implement it, Bell is developing IP caller ID.

That is all Sweldens will say though. The rest is secret.

There is a theory at Bell Labs that putting more intelligence into the network is the key to cutting the inefficiencies and restrictions of modern network security.

If Bell can succeed in making a network that can "learn and immediately react" if something is amiss, then companies will have the option to move away from the traditional firewall principle of "good inside, bad outside" to a more pervasive security, according to Sweldens.

One area of development, simply called "defence against distributed denial of service attacks", uses algorithms to help the router to "learn" what is usual traffic flow and to restrict access if an unusually high amount of traffic is coming from one source or heading towards one destination. This could be particularly advantageous against distributed denial of service attacks that strike outside the corporate firewall by flooding the access link.

Sensor networks: BT, Intel, Cisco, Xerox and others

Wireless sensor networks are causing excitement among scientists in commercial and academic R&D labs.

Sensors are small, often embedded, devices that measure a specific change in the environment, such as heat, sound, vibrations, temperature, humidity or electrical current.

The most common future scenario for the sensor network is the treatment of elderly or infirm patients in their own homes. Hundreds of sensors around the home could monitor the patient's behaviour, suggest action to the patient or report back to someone of the patient's choosing if there is an alteration from the usual pattern of events. Independently, the labs at BT and Intel are both working on such healthcare projects.

Intel's research effort is geared to adding more capability to each sensor and to adding intelligence to the network to process masses of data automatically at a local level and only pass on correct and important information or alerts.

Eric Dishman, director of proactive healthcare at Intel, explains that borrowing technology from Intel research into artificial intelligence called Bayesian networks may help to automate the way the system learns to recognise common behaviour and identifies abnormal behaviour.

Intel is also involved in pilot projects for monitoring the sub-climate (temperature and moisture) around a vineyard. But similar sensor networks could also be used to monitor the environment in office machines or stock levels in a warehouse.

In an office environment, the network could sense the presence in the meeting of a visitor's laptop computer and grant access to local devices over the wireless local area network, while blocking access to confidential information.

Quantum computing and cryptology: HP and others

The race is on to create the quantum computer, a machine able to do many simultaneous mathematical equations at lightning speed.

"Small-scale quantum devices have been demonstrated, and there is currently no fundamental reason why large-scale devices cannot be built," says Bill Munro, a senior research scientist at HP Labs in Bristol.

The problem is that quantum computers of the next decade could make child's play of the mathematical algorithms that are the basis of today's public key encryption methods, for example.

That is why HP and other research labs in the race to perfect the quantum computer are working so hard to come up with quantum-immune cryptography.

We referred these research projects to a panel of two IT directors, a venture capitalist and an analyst. Here are their comments:

On data tagging: "The concept of data tagging is a laudable one, given increasing legislation and the expanding facility for information sharing, but it will rely on all data, employees and actions being classified in some way. The thought of implementing such a mechanism is one of nightmare proportions."
Colin Simpson, group systems manager, Fuller, Smith & Turner

On sensor networks: "Perhaps George Orwell's [1984] Big Brother has an even bigger brother. There is clearly plenty to be considered from a human rights perspective in terms of how much monitoring will be done and where it will be reasonable."
Owen Williams, head of IT, Knight Frank

On ad hoc networks: "I expect to see completely dynamic ad hoc networks in two to five years, but there are issues with the routing mechanisms, and problems with interference and incompatibility between dozens of competing wireless standards. Then there is the question of the business model."
Alexander Linden, vice-president, emerging trends and technologies, Gartner Research

On IP caller ID: "This could solve a real issue for the internet. It will be tough to get all players to abide by the rules, but you could see the major players only taking traffic from sites with IP caller ID implemented. I would need to be convinced that the hackers could not break this code."
Nick Kingsbury, global sector head, software, 3i Investments

Read more on IT innovation, research and development