Why Privacy Is Critical (and Difficult) in JCAS
JCAS blends connectivity with perception. This perception is achieved by exploiting how radio energy interacts with its surroundings — people and objects reflect, absorb, and scatter the wave — and systems infer the scene from changes in amplitude, phase, timing, and direction. Essentially, a transmitter acts as a lighthouse: every transmission lights up whatever it touches. This raises immediate concerns for privacy: reflections don’t ask for consent.
Talking about privacy here isn’t about cookie banners or data breaches. Simply being within range of a JCAS system reveals something about the people present; pedestrians in instrumented streets, workers near robots, tenants in dense housing. How much is revealed depends on many things: the adversary’s vantage and signal quality, the bandwidth in use, the packet rate, how regular the signaling is, etc. That leads straight to the central tension of JCAS.
The tradeoff: performance vs. privacy
Many of the choices that make communication shine often make passive inference easier. Wider bandwidth sharpens range and material cues. Higher packet rates give an adversary more samples. Stable pilot structures and predictable schedules give bystanders the anchors needed to align and misuse. Even when payloads are perfectly encrypted, control traffic and channel estimates leave outsiders with enough information [2]. More capable links typically mean more recoverable details about the environment and its occupants — an intrinsic tradeoff JCAS systems have to make.
Design choices change the shape of risk
Sub-6 GHz reaches far and penetrates walls [1]—excellent for coverage, but risky for privacy. Some of that risk can be mitigated with transmit-side randomization: for example, introducing small pseudorandom changes in pilots or phases that only legitimate receivers, armed with shared keys or synchronization, can undo [3]. This leaves outsiders facing extra randomness, making passive inference brittle.
mmWave shifts the balance differently. Narrow beams, high path loss, and strong blockage physically constrain where energy propagates, limiting the vantage points from which an adversary can observe [6]. At the same time, mmWave systems can employ deliberate perturbations such as beam dithering or controlled hopping across beam directions [4, 5]. Legitimate receivers track these changes, but for eavesdroppers the scene appears unstable, reducing what can be inferred. The cost, however, is practical: beam training overhead, sensitivity to occlusions, and in homes the same walls that shield privacy can also deprive coverage.
What tomorrow’s JCAS should deliver
Wi-Fi sensing is useful because it lights up the scene—and that’s exactly why it’s sensitive. Over the next few years, the hard work isn’t only better models or wider bandwidths; it’s proving that presence, identity, and routine aren’t casually exposed. The bar is simple to state and hard to meet: ship features, measure leakage, and make misuse brittle by design. Future systems will then need to address privacy by design—pairing novel approaches with practical methods—and be judged against rigorous, standardized evaluation criteria to prove they deliver what they promise.
An article by Fabian Portner.
References
[1]: Adib, F., & Katabi, D. (2013). See Through Walls with Wi-Fi. Proceedings of ACM SIGCOMM 2013.
[2]: Zhu, Y., Xiao, Z., Chen, Y., Li, Z., Liu, M., Zhao, B. Y., & Zheng, H. (2020). Et Tu Alexa? When Commodity WiFi Devices Turn into Adversarial Motion Sensors. NDSS 2020.
[3]: Qiao, Y., Zhang, O., Zhou, W., Srinivasan, K., & Arora, A. (2016). PhyCloak: Obfuscating Sensing from Communication Signals. USENIX NSDI 2016.
[4]: Fan et al., “Secret-Focus: A Practical Physical-Layer Secret Communication System by Perturbing Focused Phases in Distributed Beamforming,” IEEE INFOCOM 2018.
[5]: BeamSec: A Practical mmWave Physical-Layer Security Scheme Against Strong Adversaries,
[6]: Rangan, S., Rappaport, T. S., & Erkip, E. (2014). Millimeter-Wave Cellular Wireless Networks: Potentials and Challenges. Proceedings of the IEEE, 102(3), 366–385.