Fixing the Bluetooth discovery process

Posted April 21, 2005

The idea of Bluetooth was great, but the implementation has a number of quirks making it less useful than it might have been (my personal pet peeve is the 8-device limit for a local network, which is really silly). Reportedly this was mainly due to a "too many cooks" phenomenon where everybody on the Bluetooth committee — phone people, chip people, OS people, PDA people, etc. — had their own ideas and priorities, so the final spec reflected a haphazard mix instead of a coherent, well-thought-out standard. The original spec was supposedly even contradictory in places, so it was literally impossible to make a conforming device, and early interoperability problems resulted.

One of the more painful processes is device discovery, where two phones, say, need to find each other and exchange IDs and get permission to have further communication between them. This is lengthy both in terms of fundamental communications parameters and also in terms of the user interface that phones and PDAs typically have the owners go through: for security purposes it often involves an out-of-band communication where you tell me what your phone's current passcode is (to guarantee that I'm someone you know and not just some random passerby trying to hack your address book).

This paper I ran across came up with a nice way to bypass some of this. The idea is that you take advantage of the nearly ubiquitous cameraphones to have one device display a special barcode-like glyph. The other phone takes a picture of the screen, analyzes the glyph, and from that extracts the necessary Bluetooth discovery information as well as possibly some extra info that ordinarily would need to be rolled into a subsequent communications link. Note that this process automatically incorporates the passcode sort of security step, since you need to show me your phone's screen.

The authors mainly concentrate on the case of Bluetooth-enabled vending machines, where you buy a soda with your phone (not generally available in America because our cellular system sucks). This is the sort of case where a quick discovery process for a single-use communications channel is a big win, but even in the general case, using a screen plus camera as a quick communications channel is a great idea and generalizes well to persistent glyphs (i.e. barcodes) which are also becoming common in countries with more advanced consumer technology.