Sensor networks are typically purpose-built, designed to support a single running application. As the demand for applications that can harness the capabilities of a sensor-rich environment increases, find the availability of sensing infrastructure put in place to monitor various quantities soars, there are clear benefits in a model where infrastructure can be shared amongst multiple applications. This model however introduces many challenges, mainly related to the management of the communication of the same application running on different network nodes, and the isolation of applications within the network. At the same time, security concerns related to terrorism, crime, and lower-level anti-social behaviour and fraud have placed pressure on government agencies to be seen to be doing something to respond. Extensive surveillance is the easy option, as already seen in the UK with the installation of millions of CCTV cameras and a political fondness for the "database state"[ABD~+09]. The emergence of low cost pervasive sensing will present another tempting target for surveillance. While there may be legitimate reasons for situational awareness from time to time, placing all citizens "under the microscope" 24/7 has a well-known corrosive effect on society [Fun04]. Thus, the combination of dynamic requirements for privacy, and occasional surveillance results in new security challenges. In this paper, we describe the Fresnel project's technology[ELMC10] that addresses these challenges. We tackle these by design methodologies, and integrate solutions with each sensor application, and provide a substrate that enforces appropriate levels of privacy and separation of roles and rights to data, within a virtualised sensor networked OS.
展开▼
