The term Internet of Things (IoT) is used to describe a large-scale network of electronic devices that communicates using the Internet protocol. In these networks, to enable a variety of applications, each device would exchange information about itself and its surroundings .
Sensors enable these devices to monitor environment parameters such as room temperature, navigation speed or ambient noise. The Internet of Things that is built with these capabilities would eventually become an Internet of Sensors.
The ever-growing market of RFIDs and low-cost microcontrollers introduces new types of platforms that could be used as Things. A popular example of a low-cost, passively powered platform with an integrated sensor is the Wireless Identification and Sensing Platform (WISP), designed by Intel Seattle.
In those low-cost, light-weight platforms, MEMS sensors are typically used to interact with the analog world. This interaction could enable many unique applications in a variety of fields from healthcare to cattle management.
As we build the Internet of Things, new challenges emerge concerning their security. One of the most significant challenges is being able to correctly identify each of these numerous devices. In certain critical applications, such as weapon condition monitoring for military and law enforcement, the trust that is required from device identifiers is of utmost importance.
In this paper, we target a commonly used MEMS sensor, an accelerometer, and utilize its process variations to generate digital fingerprints. This is achieved by measuring the accelerometer’s response to an applied electrostatic impulse and its inherent offset values. Our results revealed that MEMS sensors could be used as a source for digital fingerprints for run-time authentication applications.
The target platforms are often battery-limited and low-power, therefore, it is important to estimate the energy consumption of the proposed digital fingerprint generation. The target accelerometer has a scalable current consumption which is automatically tuned to reduce the power and energy cost.
The accelerometer also has a low power mode that uses a lower current consumption, but this mode is omitted in our experiments, because it introduces more noise to measured outputs . The lowest energy requirement for output data sampling is approximately 0.1 µJ and our experimental setup uses 2.52 uJ to sample one output data.
In this paper, we made an effort towards using accelerometer sensor as a source of physical authentication on low-cost platforms. We have explored the possibility of two sources, namely the self-test measurement and the offset of the uncalibrated sensors to have device-unique values due to process variation.
The uncalibrated offset measurements can be applied when the target platform is stationary and self-test measurements can be applied independent of the platform position. The proposed sources can be used as electronic fingerprints for applications that require run-time fingerprint generation without power cycling.
To read this external content in full, download the complete paper from the author archives online at Virginia Institute of Technology.