Comparison of Experimental and Theoretical Mechanical Jitter in a THz Communication Link

Abstract

The effect of mechanical vibration (jitter) is an increasingly important parameter for next-generation, long-distance wireless communication links and the channel models used for their engineering. Existing investigations of jitter effects on the terahertz (THz) backhaul channel are theoretical and derived primarily from free space optical models. These lack an empirical and validated treatment of the true statistical nature of antenna motion. We present novel experimental data which reveals that the statistical nature of mechanical jitter in 6G links is more complex than previously assumed. An unexpected multimodal distribution is discovered, which cannot be fit with the commonly cited model. These results compel the refinement of THz channel models under jitter and the resulting system performance metrics.

Figures (5)

• Figure 1: Aerial view of measurement path.
• Figure 2: Received signal showing amplitude under three different jitter levels (a) Baseline - no intentional motion (b) low jitter (c) high jitter.
• Figure 3: Beam wander at the receiver. A Gaussian beam from the Tx arrives at the Rx with radius $w_d$. The Rx radius is $a$. Azimuth and elevation deviations ($\theta_x, \theta_y)$ cause the beam center to wander in the Rx $xy$-plane with a displacement vector $r$. Only the elevation deviation $\theta_y$ is shown on the left.
• Figure 4: Analytical model from Eq. \ref{['eqn:h_m-formula']} under increasing jitter severity. The model is evaluated using the dimensions of the experimental setup. A receiver radius $a = 152.4$ mm, Gaussian transmitted beam with starting waist 152.4 mm, and transmission distance $d = 341$ m are assumed.
• Figure 5: Measured and normalized distribution of misalignment gain under increasing jitter (blue bins) and estimated analytical fits (red curves) (a) Baseline - no intentional jitter (b) low jitter (c) medium jitter (d) high jitter. The analytical model results have been overlaid on each measurement for various $\sigma_{theta}$. These curves are chosen to match the mean, variance, and highest peak of the measured PDFs.