VoltKey: Continuous Secret Key Generation Based on Power Line Noise for Zero-Involvement Pairing and Authentication
VoltKey addresses a pain point that is intrinsic to wireless networking: how to securely distribute a shared authentication key to authorized devices while restricting access to unauthorized users. Secure and usable key generation is particularly challenging for personal IoT systems because most low-cost IoT devices delegate the user interface to web- or mobile-based apps, mainly due to form factor or cost constraints. In result, commissioning and authentication of new devices tends to be a clunky process. VoltKey ameliorates this situation by transparently generating a shared common key on devices based on spatiotemporal noise extracted from the AC power line. We introduce a novel scheme to extract randomness from power line noise and address synchronization challenges resulting from using low-cost hardware design.
The hardware prototype consists of two main components: (1) the analog input circuitry for filtering and amplifying power line noise and (2) microcontroller unit which includes an analog-to-digital converter (ADC) for noise measurement and key extraction procedures. This low-cost design enables VoltKey to be easily implemented on top of standard USB power supplies as a modular addition to any IoT devices or wireless access points
SAMPLING RATE ESTIMATION
The sampling rate of the ADC between two devices has to match in order to accurately synchronize their common time. Therefore, both devices independently derives its own sampling rate, using the
60 Hz sinusoidal behavior from the power line as a common time base. We will refer to first c number of samples as a preamble period. Ideally, due to the sinusoidal nature of the power line, the index-wise average value of the equally sliced signal of length c should exhibit high correlation with the preamble period if c properly represents number of samples within each period. Above illustrates an example of comparison between a preamble period and the mean of 20 subsequent periods with different c values.
As we can see, the correlation is highest when c is 1419 and the correlation drops as c increases to 1422. Therefore, as we sweep the values of c in increasing manner, we can safely conclude that this device has 1419 samples representing each period. Once each devices goes through this process, its c value is exchanged and the device with higher value goes through downsampling procedure to agree on the number of samples per period.
KEY EXTRACTION ALGORITHM
In order to extract multiple bits, each noise period (noise component that resides in each period), is equally sliced into multiple bins, where each bin contains equal number of samples. First, device A searches for the index of the sample with the maximum absolute value among all samples in each bin for every period and sequence of the indices is shared with device B. With the common index sequence, both devices can extract the same bit sequences by observing the value of the noise at each index. If the value is greater than the mean of the noise period, a bit 1 is extracted from the bin of the period; otherwise, bit 0 is extracted.
To validate that the superimposed noise signal is spatially unique, as a function of the distance between two authenticating devices, we set four VoltKey (A, B, C, D and E) Devices to authenticate with a single access point (A) under varying distance within a realistic laboratory environment under regular daily usage. When the two devices are 1.0 m apart, the bit agreement rate of 128-bit key is above 93.5% regardless of number of bits harvested within each noise period. As distance increases up to 24.8 m, high bit agreement rate of above 92% is maintained. One other observation is that when distance between authenticating devices increases, noise in the power line is translated into different bits, which proves the spatial uniqueness property of our key generation algorithm.
Realistic Deployment Scenario
To verify the overall effectiveness of VoltKey under realistic deployment scenarios, we set four devices within a regular daily environment of one-bedroom apartment and office to measure the the bit agreement rate of each device. The bit agreement rate for device deployed in office environment exhibit higher rate compared to that of a one-bedroom apartment, due to that the greater number of switching activities of electronic appliances in the apartment leading to higher fluctuation on the voltage signal. Overall, high bit agreement rate of above 92% can be expected to be maintained for all devices in both one-bedroom and office scenarios.
Kyuin Lee, Neil Klingensmith, Suman Banerjee, and Younghyun Kim, "VoltKey: Continuous Secret Key Generation based on Power Line Noise for Zero-Involvement Pairing and Authentication," in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), Vol. 3, No. 3, pp. 93:1–93:26, 2019 (Presented at the ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) 2019)
[PDF] [Presentation] [ACM link]