Wireless Ontologies - 2010

Tim McFadden

This note discusses middleware for making wireless (RF) devices more practical and powerful in typical wireless environments. In the vision spirit, let’s look five years from now to 2010, on the skin of cyberspace, away from stationary Internet connections, where wireless lives.

Most handhelds don’t know what’s going on around them; they don’t have context-awareness. Computer ontologies can provide an open industry standard for delivering such data; they define what computers can know in different situations. Wireless ontologies deliver context-awareness to wireless devices, to let them live in many different situations.

Resource Description Framework (RDF) and Knowledge Interchange Format (KIF) are examples of ontology languages. RDF is expanding in usage with the semantic web and the Liberty Alliance Project has its own corresponding meta-data protocol. Ontologies can help realize the dream of Tim Berners-Lee, of allowing an agent on one side of the web to describe any object to an agent on the other side of the web. Our world makes this goal quite elusive, of course, but the path is a really useful one.

The raw, uncontrolled field environment of mobile wireless devices means that they must meet many different situations. Given the small size of wireless devices, compared with commodity PCs, wireless ontologies are critical for changing devices’ behavior to be more useful in each different situation. This is as opposed to having different methods for each situation, which can get very expensive. A smaller set of software can be powered by changing context-awareness.

For example, “If you’re not real-time you’re dead.” A supply-chain wireless ontology can enable one person with a cell phone/headset/computer/display to walk onto many different sites and get the picture. Distributed supply chains have not yet been mastered by most companies.

For further example, “Killer apps need killer data.” Computers are stupid and irritating, at least according to 90% of the users. Computers need to be told what to do and what’s happening. Wireless ontologies allow humans to describe the everyday situations that so confuse computers and there are lots of everyday situations. This is especially critical, because AI only works in small, closed worlds and even then it’s very time consuming to produce. Wireless devices work in the above mentioned raw, uncontrolled field environment, which is right in the tar pits of AI. Of course software agents are still mostly AI, but they are a very useful human /computer model.

For further example, the killer data needed is often real-time data, which make RF tags look good.

RF tags, e.g. RFID, are already a growing industry. RF tags can just be used to simply identify objects to servers (RFID) or as smart tags that can actively organize themselves into federations that can track collections of containers or as smart tags that can squirt real-time and built-on-location data. I headed a project at France Telecom R&D, California, that had smart RF tags squirting real-time RDF to an iPAQ, which then did WiFi to a server (Java tech, of course). A collection of the smart tags would organize into an ad hoc network and report back to the iPAQ in real-time.

What then can we ship on January 2nd 2010? - wireless devices that can go from construction site to street to factory, being useful in each situation, picking up data from tags on freight cars, automobiles, boxes of wine, or bus stops. The user will see bills of materials, routes, temperatures, or directions. Fuzzy agents will be a critical technology, of course, but that’s on another page.