Separating raw sensor data from processed inferences

The sticky issue of who gets sensor data has been addressed by Guruduth Banavar and Abraham Bernstein in “Challenges in Design and Software Infrastructure for Ubiquitous Computing Applications” in the book Advances in Computers, Vol. 62, parts of which you can view at Amazon or Google Books:

Gathering data of any kind irrevocably leads to privacy concerns. Where should the data be stored and what boundaries shouldn’t it cross? Who should have access and who doesn’t? These questions aren’t new to ubiquitous computing. But the pervasiveness of these sensors adds a new layer of complexity to understanding and managing all the possible data streams. Can one subpoena the data collected by ubiquitous computing systems? As the answer is probably yes, there might be a demand for ubiquitous computing systems where the raw sensor data cannot be accessed at all, but only processed inferences from the data, like “burglar entry,” can.

Quite right, there is such a demand. How do we move forward from the demand to the reality? —Chris Peterson

One Response to “Separating raw sensor data from processed inferences”

  1. subramanian says:


    my concern is the still poor match between what the terms used for easy human interface (walls) and what they mean when technically translated for operational reasons. there are no walls, it is only a term used to relate a concept to the raw individual. But there exists a huge gap between this translation and inference drawn from fine grained digital data and its meaning drawn by the lay man

Leave a Reply