A Measurement-Based Spatially Consistent Channel Model for Distributed MIMO in Industrial Environments

Abstract

Future wireless communication systems are envisioned to support ultra-reliable and low-latency communication (URLLC), which will enable new applications such as compute offloading, wireless real-time control, and reliable monitoring. Distributed multiple-input multiple-output (D-MIMO) is one of the most promising technologies for delivering URLLC. This paper classifies obstructions and derives a channel model from a D-MIMO measurement campaign carried out at a carrier frequency of 3.75 GHz with a bandwidth of 35 MHz using twelve fully coherent distributed dipole antennas in an industrial environment. Channel characteristics are investigated, including statistical measures such as small-scale fading, large-scale fading, delay spread, and transition rates between line-of-sight and obstructed line-of-sight conditions for the different antenna elements, laying the foundations for an accurate channel model for D-MIMO systems in industrial environments. Furthermore, to ensure spatial consistent simulation results the correlations of large-scale fading between antennas are modeled using Gaussian random fields. Lastly, tail distributions are included to enable proper evaluations of reliability and rare events. Based on the results, a channel model for D-MIMO in industrial environments is presented together with a recipe for its implementation.

Figures (20)

• Figure 1: The rich scattering and heavily shadowed industrial environment where the D- measurements were conducted. The dots mark the approximate locations of the antennas.
• Figure 2: Top-down overview of the industrial environment where the D- measurements were conducted. The photograph in Fig. \ref{['fig:photo-hall']} is taken from the position indicated by the eye. The anchors are visible along the two sides and the trajectories driven by the agent are visualized. The numerology of the anchors are shown for anchors 1, 6, 7, and 12; the rest are implicit for visual purposes. The solid and dashed lines represent different measurement runs, i.e. measurements captured during different times.
• Figure 3: The amount of obstruction of the (approximated) first Fresnel ellipse, over time for all anchors. The scenario depicted is the loop scenario.
• Figure 4: Left: The distribution of the anchor and agent separation .Right: the of the approximated obstruction of the first Fresnel ellipse. From the of the obstruction it is clear that the environment in which the data has been collected is challenging, with a lot of obstructing machinery between the agent and the anchors. The shows a 100 % obstruction in more than 30 % of the collected data, given the choice of a $2\lambda$ radii of the cylinder.
• Figure 5: The of the number of links with LoS conditions throughout the measurements.