The Inference Problem and Pervasive Computing

The promise of pervasive computing systems, which flood an environment with sensors and smart devices, has been diminished by an inability to adequately address privacy concerns. Measurements of privacy demonstrate a high level of public concern, yet designers of pervasive computing systems do not find these measurements relevant to the use of their systems. This has led designers to distrust reports of privacy unease. A potential solution to this impasse may be found in a privacy condition from the data mining literature known as the ―inference problem.‖ The inference problem occurs in data mining when confidential information can be derived by released data. Privacy issues within pervasive computing systems are most acute when they involve a variation of the inference problem. This paper will discuss methods used in data mining to limit the inference problem, and discuss their application to pervasive computing. Finally, a research agenda will suggest how to identify instances of the inference problem within pervasive computing systems.