Appropriate interaction structure for wireless deployment utility representation

Determine which interaction structure—individual, pairwise, or higher-order relational modeling—provides the appropriate reward representation for multi-transmitter wireless deployment under different urban environments and deployment scales.

Background

CityDeploy-Bench models multi-transmitter deployment utility as a property of the complete transmitter set because coverage complementarity, redundancy, serving-cell reassignment, and co-channel interference create dependencies among deployed transmitters. The paper compares pooled, pairwise, latent-relational, and higher-order hypergraph utility representations.

The authors report that pairwise or latent-relational models can be competitive at low transmitter cardinalities, whereas higher-order structure becomes increasingly useful in strongly coupled settings. However, the appropriate relational order may vary with urban geometry, deployment scale, and physical coupling, so the general choice of reward structure remains unresolved.

References

The open question is therefore not simply whether relational modeling improves planning, but which interaction structure provides the appropriate reward representation under different urban environments and deployment scales.