In mmWave communications, channel is of the utmost importance. In fact, it presents significant differences from the classical sub-6 GHz wireless channel, some of them trivial (such as an increased path loss), some of them more complex and characteristic (e.g., channel sparsity, more prominent diffuse rays due to increased relative surface roughness, harsher blockage).

Channel models are generally subdivided into three categories:

  • Analytic: they are the simplest models, with desirable properties for analytical derivations but, for this reason, the least accurate. Examples are Rayleigh/Rician/Nakagami-m models regarding fading, and simple sectors for beamforming often in 2D
  • Stochastic: they tend to be popular in large scale simulations given their relative speed and level of realism. Still, there is great debate in channel model community whether these models are accurate enough and, especially, how to correlate them meaningfully over space and time. Examples are the 3GPP TR 38.901 and NYUSIM
  • (Quasi-)Deterministic: Typically based on ray-tracing, they are by far the most accurate though complex channel models.

An implementation of an open-source ray-tracer for mmWave was created by our group in collaboration with NIST and can be found here.
Output traces obtained by it are currently supported by the full stack network-level simulator ns-3 such as the IEEE 802.11ad module.
Some results can be seen here.

Title  DateArea
P. Testolina, M. Lecci, M. Polese, M. Giordani, M. Zorzi, "Scalable and Accurate Modeling of the Millimeter Wave Channel", IEEE International Conference on Computing, Networking and Communications (ICNC), Big Island, HI, 2020
arXiv
2020/02Channel Modeling
M. Lecci, P. Testolina, M. Giordani, M. Polese, T. Ropitault, C. Gentile, N. Varshney, A. Bodi, M. Zorzi, "Simplified Ray Tracing for the Millimeter Wave Channel: A Performance Evaluation," Information Theory and Applications Workshop (ITA), San Diego, US, 2020
arXiv
2020/02Channel Modeling