Comparison with blockage-aware learned observation-to-beam baselines

Compare WM4ISAC with learned observation-to-beam baselines specifically trained to handle dynamic blockage and select non-line-of-sight backup beams.

Background

The paper evaluates WM4ISAC against current line-of-sight beam tracking, a perfect non-line-of-sight beam reference, and a perfect-information upper bound. It notes that existing learning-based predictive beamforming methods primarily predict the next line-of-sight beam and are not directly applicable when the user echo disappears during blockage and a non-line-of-sight backup beam must be selected.

A direct comparison with learned observation-to-beam methods designed for blockage-aware beam selection is not conducted in the paper and is explicitly deferred to future work. Such a comparison would assess WM4ISAC against learning-based baselines that address the same blockage-resilient beam-management setting.

References

Learning-based predictive beamforming predicts the next LoS beam and is not directly applicable here, since the user echo vanishes during blockage and no NLoS backup beam is selected; a comparison with learned observation-to-beam baselines trained for blockage is left for future work.

— WM4ISAC: World Model for Proactive ISAC Under Dynamic Blockage  (2610.08267 - Xiao et al., 6 Oct 2026) in Section IV-A, Online Rate Evaluation