HackMan: hacking commodity millimeter-wave hardware for a measurement study

The extremely high frequency of Millimeter-Wave technology warrants Gbps throughput for the next-generation wireless communication systems, but mmWave signals also suffer from severe path loss due to high attenuation. To compensate for this loss, mmWave radios establish communication links via directional beams so as to increase channel gains and communication range. Until recently, the measurement studies on mmWave technology were mainly based on prototypes built from band-limited Software Defined Radio, which could not characterize performance in realistic settings. Latest studies using commodity hardware reports straightforward measurements on the impact of environment settings but did not present deep analysis on the correlations of low-layer information. Meanwhile, those studies lack the ability to configure commodity devices under controlled settings, for instance, a single beam pattern for experimentation, thus failing to perform deeper analysis on low-layer protocol parameters. In this paper, we conduct extensive measurements in typical indoor settings, utilizing 802.11ad-compliant commodity hardware. Different from earlier studies, we hack the firmware and gain the privilege to modify physical layer settings online, enabling us to gain more insights under controlled settings. Essentially, we have demonstrated that (1) Signal-to-Noise Ratio, the criteria for beam control, may not be positively correlated with throughput, (2) sticking to a single beam pattern during data transmission can lead to both channel gains and throughput improvement, and (3) only independent cross-links could interfere with each other while multi-links coordinated by one AP experience no interference. These insights lead us to rethink the existing beam control policy.