BROWSE
Detail
Journal Article
Exhaustive Structure Learning of R-Vine Copula Models for Analyzing Multiband Spectrum Utilization in Mobile Communications
- Authors
- Yunbae Kim, Jonghun Yoon, Hyeyeon Kwon, Seungkeun Park
- Issue Date
- 2026-03
- Citation
- IEEE Access, v.14, pp.33460-33476
- ISSN
- 2169-3536
- Publisher
- IEEE
- Language
- English
- Type
- Journal Article
- Abstract
- Accurate modeling of multiband spectrum utilization is essential for data-driven spectrum management and refarming in mobile communication systems. However, the bounded and non-Gaussian nature of Resource Block (RB) utilization, together with complex inter-band dependencies, challenges conventional multivariate modeling approaches. This work proposes a flexible density estimation framework based on Regular vine (R-vine) copulas, which construct joint distributions from bivariate components. Since practical LTE and 5G deployments involve only a small number of concurrently used bands, we perform an exhaustive search over all admissible R-vine structures to identify the optimal dependence configuration. To support this search, we develop an efficient enumeration algorithm that incrementally constructs valid edge sets while directly enforcing the proximity condition. For each pair-copula term, we adopt Gaussian Mixture Copula Models to capture complex and nonparametric dependence patterns, and we derive gradient expressions that enable efficient likelihood-based parameter estimation. The proposed framework is validated using real-world LTE measurements of multiband RB utilization. Goodness-of-fit evaluations confirm that the estimated distributions accurately reflect observed patterns, while the learned dependence structures reveal interpretable inter-band behaviors, preferential usage relationships, and band-specific load characteristics. These results demonstrate the suitability of flexible copula-based modeling, particularly in low-dimensional settings relevant to simulation-driven spectrum planning in modern communication systems.
- Keyword
- Gaussian mixture copula models, mobile communication networks, multivariate density estimation, spectrum utilization analysis, vine copulas
- KSP Keywords
- Accurate modeling, Data-Driven, Density estimation, Dependence structures, Efficient enumeration, Enumeration algorithm, Gaussian mixture(GM), Goodness-of-Fit, Inter-band, Load Characteristics, Mobile communication networks
This work is distributed under the term of Creative Commons License (CCL)
(CC BY NC ND)
(CC BY NC ND)
