I’m writing a paper at the moment with the provisional title A Context-Sensitive Service Discovery Protocol for Mobile Computing Environments. It is to be submitted to the Mobile Computing and Communications Review (MC2R). In that paper I describe an approach for
- enabling queries and advertisements to be context-sensitive, and
- allowing applications to be context-aware by issuing context-sensitive queries.
These goals are achieved by enabling sub-queries to stand in for actual query components. These sub-queries differ from those of say, SQL, in that the author of the query needn’t be fully aware of the structure of that sub-query. Rather, the sub-query can be specified in very general, abstract terms, and the protocol does the rest. For instance, if a user needs to issue a query sensitive to the current location, except that she is not aware of her location, she creates a query for the type of service she’s looking for but leaves the location components for the service discovery infrastructure to fill in. The infrastructure can fill in the location information by any means available to it. A poor man’s solution is to fill in the missing components using approximate location information which can be provided by the resolver that receives the query (i.e. the resolver, because it’s fixed furniture, can be preconfigured with its own location and use this to augment any location-sensitive queries it receives). Alternatively, and increasingly likely because of the headway being made into location services research, a location manager can announce the user’s location as an advertisement every time the user’s location changes (i.e. if the user moves from one room to another). This advertisement can then be used to complete the context-sensitive query. Note that advertisements can also be context-sensitive. In addition, query results are ranked by a preference mechanism, and the query relaxation function (called scoping in Superstring) enable some neat things to be done. Furthermore queries can be made persistent, so that they become rather like subscriptions in a content-based messaging protocol such as Elvin. I’m not going to elaborate on this further because I’ve spent the best part of the last week writing about it. Some more later, perhaps.
I had a lovely weekend. Saturday consisted of cleaning, eating lunch at Sherwood Forest Park, shopping and watching Benny and Joon. Sunday was very easy going. So easy going that I don’t really remember what I did, except that it involved reading some papers and a couple of articles from New Scientist. Stephen Hawking has done a back flip with respect to his prior view that no information can escape from a black hole. Some guy in Taiwan has conducted a study that shows that people get less frustrated when computers apologise for being stupid (or at least when they don’t give terse feedback when something goes wrong). Another article says that it’s not necessarily a good thing to be intelligent. I’ve been trying to tell my intelligent friends that for years, so this is great news! ;). It seems likely that Poincaré’s conjecture (to do with topologies, and which I’m not even going to pretend to understand) has been solved by a Russian mathematician named Perelman. This isn’t exactly recent news, but it’s taking a long time for his work to be peer reviewed.
Nigel and I are going to Sydney this weekend for the hell of it. I hope to see my cousins and MM. Actually I’ll be seeing MM on Saturday, but haven’t yet organised to see the cousins. Better call them right now…
