Exploiting Randomized Continuous Wave in Secure Backscatter Communications

To enable the low-cost ubiquitous Internet of Things, passive backscatter communication is envisioned as one of the most prominent and promising techniques; however, the underlying security issues associated with practical finite-alphabet signaling from the perspective of physical-layer security (PLS) have not been well studied. Despite several preliminary efforts on improving the eavesdropper's decoding error probability through PLS approaches, this article comprehensively investigates the secrecy rate performance of a secure multiantenna radio-frequency identification (RFID) system with a finite-alphabet input at the RFID tag. Unlike conventional noise-injection schemes, a randomized continuous wave (CW) signal is exploited at the RFID reader for security enhancement, and an analytical framework is proposed to evaluate the impact of exploiting either full or only statistical knowledge of the randomized CW signal at the reader and the eavesdropper, respectively. The secrecy rate is maximized by designing the transmitted randomized CW signal to tackle the stability-variance tradeoff between balancing legitimate signal reception and eavesdropper mitigation. In particular, we show that the proposed scheme also poses a tradeoff between the received additive and multiplicative noise at the eavesdropper for the special case of a single-antenna eavesdropper. Moreover, the more practical case where the eavesdropper's instantaneous channel state information is unavailable is studied under different fading conditions. The numerical results verify the accuracy of the proposed approximations and show that introducing a small variance into the CW signal can greatly improve the system secrecy.