Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Channel Knowledge Map for Cellular-Connected UAV via Binary Bayesian Filtering

Yuhang Yang, Xiaoli Xu, Yong Zeng, Haijian Sun, Rose Qingyang Hu

TL;DR

The paper tackles the challenge of building a location-specific LoS probability map for cellular-connected UAVs by treating the link state as a binary spatial field $l(\mathbf{x})$ and initializing it with an empirical elevation-based LoS model. It then introduces a binary Bayesian filter that sequentially updates the LSM using radio measurements, augmented by explicit spatial correlations in distance and azimuth to propagate information to unmeasured locations. The authors derive a practical update rule for the log odds and an exponential azimuth correlation model based on the phi coefficient, enabling two-step LSM refinement that leverages measurements where available and nearby correlated directions otherwise. Results in a simulated urban environment show that the proposed method substantially improves LSM accuracy over priors and KNN baselines, particularly under sparse measurement scenarios, underscoring its potential for environment-aware CKM assistance in UAV communications and sensing.

Abstract

Channel knowledge map (CKM) is a promising technology to enable environment-aware wireless communications and sensing. Link state map (LSM) is one particular type of CKM that aims to learn the location-specific line-of-sight (LoS) link probability between the transmitter and the receiver at all possible locations, which provides the prior information to enhance the communication quality of dynamic networks. This paper investigates the LSM construction for cellularconnected unmanned aerial vehicles (UAVs) by utilizing both the expert empirical mathematical model and the measurement data. Specifically, we first model the LSM as a binary spatial random field and its initial distribution is obtained by the empirical model. Then we propose an effective binary Bayesian filter to sequentially update the LSM by using the channel measurement. To efficiently update the LSM, we establish the spatial correlation models of LoS probability on the location pairs in both the distance and angular domains, which are adopted in the Bayesian filter for updating the probabilities at locations without measurements. Simulation results demonstrate the effectiveness of the proposed algorithm for LSM construction, which significantly outperforms the benchmark scheme, especially when the measurements are sparse.

Channel Knowledge Map for Cellular-Connected UAV via Binary Bayesian Filtering

TL;DR

The paper tackles the challenge of building a location-specific LoS probability map for cellular-connected UAVs by treating the link state as a binary spatial field and initializing it with an empirical elevation-based LoS model. It then introduces a binary Bayesian filter that sequentially updates the LSM using radio measurements, augmented by explicit spatial correlations in distance and azimuth to propagate information to unmeasured locations. The authors derive a practical update rule for the log odds and an exponential azimuth correlation model based on the phi coefficient, enabling two-step LSM refinement that leverages measurements where available and nearby correlated directions otherwise. Results in a simulated urban environment show that the proposed method substantially improves LSM accuracy over priors and KNN baselines, particularly under sparse measurement scenarios, underscoring its potential for environment-aware CKM assistance in UAV communications and sensing.

Abstract

Channel knowledge map (CKM) is a promising technology to enable environment-aware wireless communications and sensing. Link state map (LSM) is one particular type of CKM that aims to learn the location-specific line-of-sight (LoS) link probability between the transmitter and the receiver at all possible locations, which provides the prior information to enhance the communication quality of dynamic networks. This paper investigates the LSM construction for cellularconnected unmanned aerial vehicles (UAVs) by utilizing both the expert empirical mathematical model and the measurement data. Specifically, we first model the LSM as a binary spatial random field and its initial distribution is obtained by the empirical model. Then we propose an effective binary Bayesian filter to sequentially update the LSM by using the channel measurement. To efficiently update the LSM, we establish the spatial correlation models of LoS probability on the location pairs in both the distance and angular domains, which are adopted in the Bayesian filter for updating the probabilities at locations without measurements. Simulation results demonstrate the effectiveness of the proposed algorithm for LSM construction, which significantly outperforms the benchmark scheme, especially when the measurements are sparse.
Paper Structure (14 sections, 1 theorem, 37 equations, 14 figures, 1 table, 2 algorithms)

This paper contains 14 sections, 1 theorem, 37 equations, 14 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. System model
  3. Measurement model
  4. An Overview of Binary Bayesian Filter
  5. LSM Construction
  6. Link State Mapping based on Radio Measurements
  7. Probabilistic LSM update based on binary Bayesian filter
  8. The spatial correlation of LoS link probability
  9. LSM Construction Algorithm
  10. Numeric Results
  11. Conclusion
  12. Binary Bayesian filter for LSM mapping
  13. The derivation of inverse measurement model $\Pr(l(\mathbf{x})|z_n)$
  14. The derivation of spatial correlation $r_{ij}(\mathbf{x},\mathbf{x}_n)$

Key Result

Lemma 1

After receiving the $n$-th radio measurement $z_n$, the logarithmic probability ratio $\mathcal{L}_{n} (\mathbf{x})$ can be obtained as where $\Pr(l(\mathbf{x})=1|z_n)$ is the posterior LoS probability at $\mathbf{x}$ and $k(\mathbf{x},z_n)$ is the posterior probability ratio at $\mathbf{x}$ based on the single measurement $z_n$.

Figures (14)

  • Figure 1: An overview of the LSM construction methods for cellular-connected UAVs.
  • Figure 2: LSM construction for cellular-connected UAVs.
  • Figure 3: The distribution of channel measurements conditioned on the LoS link status.
  • Figure 4: The LSM spatial correlation in different locations. For example, if point C is NLoS, then point D, which is at the same angle with respect to point O and behind C, must also be NLoS. If point A is LoS, then point B, at the same angle with respect to point O and between A and O, must also be LoS. Point O is the projection of GBS antenna on the $\mathcal{X}_h$.
  • Figure 5: The spatial correlation curve $\rho - \Delta \phi$ corresponding to different $\beta$.
  • ...and 9 more figures

Theorems & Definitions (2)

  • Lemma 1
  • proof