Empirical blockage characterization and detection in indoor sub-THz communications

• Published in: Computer communications 2023-03, Vol.201, p.48-58
• Main Authors: Shurakov, Alexander, Moltchanov, Dmitri, Prikhodko, Anatoliy, Khakimov, Abdukodir, Mokrov, Evgeny, Begishev, Vyacheslav, Belikov, Ivan, Koucheryavy, Yevgeni, Gol’tsman, Gregory
• Format: Article
• Language: English
• Subjects: Blockage detection algorithm; Blockage metrics; Indoor deployment; Sub-THz communication
• ISSN: 0140-3664; 1873-703X
• DOI: 10.1016/j.comcom.2023.01.017

The next step in the last mile wireless access is utilization of the terahertz (THz) frequency band spanning from 0.1 to 3 THz, specifically, its lower part (up to 300 GHz) also known as sub-THz frequencies. At these frequencies, communication systems can offer tens of consecutive gigahertz potentially allowing to further improve access rates at the air interface to dozens of gigabits-per-second. However, the effect of blockage evident already at millimeter waves is expected to be much more pronounced at frequencies beyond 100 GHz. In this paper, we empirically investigate the characteristics of the human body blockage by conducting a measurement campaign at carrier frequency of 156 GHz in the indoor environment. We concentrate on both mean attenuation and time-related metrics of the blockage process including the signal fade and rise times, and blockage duration. For a point-to-point transmission over a distance of 3–7 m, we find that the mean attenuation is in the range of 8–15 dB depending on the line-of-sight (LoS) height and the transmitter-to-receiver (Tx-to-Rx) distance. The blockage duration varies within 5%–10% for different Tx-to-Rx distances (with corresponding nominal values of 360–390 ms) while the signal rise and fall times gradually increase from 60 to 100 ms with the Tx-to-Rx distance growth and remain unchanged for different LoS heights. The developed blockage detection algorithm allows one to identify the blockage occurrence with a probability of 0.96–0.98 within 1–3 ms at the channel sampling rate of 500 ksample/s and 3–5 event/s of false alarm rate which is on par with modern machine learning based approaches.