Cross Body Signal Pairing (CBSP): A Key Generation Protocol for Pairing Wearable Devices with Cardiac and Respiratory Sensors

In this paper, we propose a cross body signal pairing (CBSPCR) protocol to enable key generation between wearable devices with cardiac and respiratory sensors. Most previous pairing protocols require both wearable devices to access the same biometric signal for pairing. The restriction can greatly limit communication between wearable devices and the development of new applications on the devices. We implemented CBSPCR for pairing wearable devices based on respiratory and cardiac signals. Two pairing modes are proposed for the pairing: (1) The model-based pairing is based on mathematical models between respiratory and cardiac signals. (2) The extraction-based pairing is based on the extraction of respiratory signals from cardiac signals. The pairing performance of these two modes varies with exercise intensity. To take full advantage of both pairing modes, we developed a mode switching based on the respiration rate, a reliable indicator of exercise intensity level. CBSPCR uses Lloyd-Max quantization and a BCH-based mismatch correction method to digitize the signals and correct mismatches between bit-strings. Our extensive experiments with 30 participants show that CBSPCR can generate about 0.0125 keys per second with high resistance to impersonation attacks and negligible battery consumption.