A Scalable Cloud-Edge Collaborative CKM Construction Framework Enabled by a Foundation Prior Model

TL;DR

This paper tackles scalable CKM construction for AI-native 6G by decoupling a general CKM prior from task-specific observations. It introduces a cloud-trained foundation score model to learn an unlabeled CKM prior and an edge-side plug-and-play posterior inference framework that combines this prior with local observations via a configurable likelihood, enabling zero-shot adaptation across tasks and devices. The approach, evaluated on CKMImageNet, achieves competitive construction accuracy while dramatically reducing training data needs and avoiding negative transfer, demonstrating improved deployment scalability in cloud--edge AI-RAN settings.

Abstract

Channel knowledge maps (CKMs) provide a site-specific, location-indexed knowledge base that supports environment-aware communications and sensing in 6G networks. In practical deployments, CKM observations are often noisy and irregular due to coverage-induced sparsity and hardware-induced linear/nonlinear degradations. Conventional end-to-end algorithms couple CKM prior information with task- and device-specific observations, and require labeled data and separate training for each construction configuration, which is expensive and therefore incompatible with scalable edge deployments. Motivated by the trends toward cloud-edge collaboration and the Artificial Intelligence - Radio Access Network (AI-RAN) paradigm, we develop a cloud-edge collaborative framework for scalable CKM construction, which enables knowledge sharing across tasks, devices, and regions by explicitly decoupling a generalizable CKM prior from the information provided by local observations. A foundation model is trained once in the cloud using unlabeled data to learn a generalizable CKM prior. During inference, edge nodes combine the shared prior with local observations. Experiments on the CKMImageNet dataset show that the proposed method achieves competitive construction accuracy while substantially reducing training cost and data requirements, mitigating negative transfer, and offering clear advantages in generalization and deployment scalability.

Figures (11)

• Figure 1: An illustration of cloud--edge knowledge sharing for CKM construction.
• Figure 2: Proposed cloud--edge collaborative knowledge-sharing CKM construction framework.
• Figure 3: Score-based CKM construction, by solving the reverse SDE to draw a sample from $p(\mathbf{x})$ after learning the score $\nabla_{\mathbf{x}_{i}}\log p(\mathbf{x}_{i})$ at every timestep $i$.
• Figure 4: CKM construction guided by observation $\mathbf{y}$ that draws a sample from $p(\mathbf{x}|\mathbf{y})$. The prior score $\nabla_{\mathbf{x}_{i}}\log p(\mathbf{x}_{i})$ is replaced with the posterior score $\nabla_{\mathbf{x}_{i}}\log p(\mathbf{x}_{i}|\mathbf{y})$.
• Figure 5: Comparison of the AoA map and the AoA-sine map.