Millimeter-wave (mm-wave) communication systems achieve extremely high data rates and provide interference-free transmissions. to overcome high attenuations, they employ directional antennas that focus their energy in the intended direction. Transmissions can be steered such that signals only propagate within a specific area-of-interest. Although these advantages are well-known, they are not yet available in practical networks. IEEE 802.11ad, the recent standard for communications in the unlicensed 60 GHz band, exploits a subset of the directional propagation effects only. Despite the large available spectrum, it does not outperform other developments in the prevalent sub-6 GHz bands. This underutilization of directional communications causes unnecessary performance limitations and leaves a false sense of security. For example, standard compliant beam training is very time consuming. It uses suboptimal beam patterns, and is unprotected against malicious behaviors. Furthermore, no suitable research platform exists to validate protocols in realistic environments. To address these challenges, we develop a holistic evaluation framework and enhance the performance and security in practical mm-wave communication systems.

Besides signal propagation analyses and environment simulations, our framework enables practical testbed experiments with off-the-shelf devices. We provide full access to a tri-band router’s operating system, modify the beam training operation in the Wi-Fi firmware, and create arbitrary beam patterns with the integrated antenna array. This novel approach allows us to implement custom algorithms such as a compressive sector selection that reduces the beam training overhead by a factor of 2.3. By aligning the receive beam, our adaptive beam switching algorithm mitigates interference from lateral directions and achieves throughput gains of up to 60%. With adaptive beam optimization, we estimate the current channel conditions and generate directional beams that implicitly exploit potential reflections in the environment. These beams increase the received signal strength by about 4.4 dB.

While intercepting a directional link is assumed to be challenging, our experimental studies show that reflections on small-scale objects are sufficient to enable eavesdropping from afar. Additionally, we practically demonstrate that injecting forged feedback in the beam training enables Man-in-the Middle attacks. With only 7.3% overhead, our authentication scheme protects against this beam stealing and enforces responses to be only accepted from legitimate devices.

By making beam training more efficient, effective, and reliable, our contributions finally enable practical applications of highly directional transmissions.