Channel Knowledge Map Construction: Recent Advances and Open Challenges
Zixiang Ren, Juncong Zhou, Jie Xu, Ling Qiu, Yong Zeng, Han Hu, Juyong Zhang, Rui Zhang · 2025
Read Channel Knowledge Map Construction: Recent Advances and Open Challenges by Zixiang Ren, Juncong Zhou, Jie Xu, Ling Qiu, Yong Zeng, Han Hu, Juyong Zhang, Rui Zhang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-04944.txt.
Introduction
Zixiang Ren and co-authors survey how to build a channel knowledge map, a map that stores what a radio channel is like at each place. Such maps help wireless nets plan ahead and tune links for place-aware service and sensing.
The survey first looks at classic gap-filling methods and their limits in real use. It then turns to image tools and generative AI, and to new wireless radiance field methods that build sharp maps from sparse tests. The close lists open needs, such as fast updates, use across domains, and low-cost rollout.
Catalog record
| CREATOR | Zixiang Ren, Juncong Zhou, Jie Xu, Ling Qiu, Yong Zeng, Han Hu, Juyong Zhang, Rui Zhang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-04944
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Channel Knowledge Map Construction: Recent Advances and Open Challenges Zixiang Ren, Juncong Zhou, Jie Xu, Ling Qiu, Yong Zeng, Han Hu, Juyong Zhang, Rui Zhang Abstract Channel knowledge map (CKM) has emerged as a pivotal technology for environment-aware wireless communications and sensing, which provides a priori location-specific channel knowledge to facilitate network optimization. Efficient CKM construction is an important technical problem for its effective implementation. This article provides a comprehensive overview of recent advances in CKM construction. First, we examine classical interpolation-based CKM construction methods, highlighting their limitations in practical deployments. Next, we explore image processing and generative artificial intelligence (AI) techniques, which leverage feature extraction to construct CKMs based on environmental knowledge. Furthermore, we present emerging wireless radiance field (WRF) frameworks that exploit neural radiance fields or Gaussian splatting to construct high-fidelity CKMs from sparse measurement data. Finally, we outline various future research directions in real-time and cross-domain CKM construction, as well as cost-efficient deployment of CKMs. I Introduction Channel knowledge map (CKM) has emerged as a cornerstone technology for sixth-generation (6G) wireless networks, enabling a paradigm shift from environment-ignorant network design to proactive and intelligent environment-aware optimization [ 1 ] . In general, CKM is a representation framework that provides the functional mapping from the spatial positions of transceivers to environment-specific channel knowledge such as channel gains, channel state information (CSI), reference signal strength indicator (RSSI), and multi-path delays/Dopplers/angles-of-arrival (AoAs)/angles-of-departure (AoDs) [ 2 , 3 ] . Depending on application scenarios, CKMs can be represented in various forms, like spatial databases, images, and even neural network models. By providing location- and environment-specific channel knowledge a priori , CKM has emerged as an effective tool to facilitate channel prediction, enable predictive wireless communications, and support environment-aware beam tracking and network deployment optimization [ 4 , 5 ] . CKMs can be classified into base station (BS)-to-any (B2X) and any-to-any (X2X) types according to the input dimension of transceiver positions. B2X CKM provides channel knowledge associated with a fixed-position BS to support BS-centric communications, in which the input may correspond to the two-dimensional (2D) user position in, e.g., terrestrial networks with all users at the same altitude or three-dimensional (3D) positions of users for, e.g., low-altitude networks with users distributed in 3D space. By contrast, X2X CKM takes both transmitter (Tx) and receiver (Rx) positions as input, allowing the construction of channel knowledge between arbitrary Tx-Rx pairs. Depending on whether 2D or 3D Tx/Rx positions are considered, the X2X CKM can be constructed in 4D or 6D formats, respectively. Among them, 6D X2X CKM offers the most comprehensive channel knowledge, which can thus be employed for complex tasks such as the predictive resource allocation for device-to-device (D2D) communications, unmanned aerial vehicle (UAV) trajectory planning, and environment-aware network deployment. The construction and utilization of CKMs are two stages for the realization of environment-aware wireless communication and sensing, and the successful construction of accurate CKM is of foremost significance [ 5 ] . CKM construction refers to the process of generating a spatially resolved representation of channel knowledge by leveraging prior measurement data and/or physical environmental information. For example, ray tracing is a commonly adopted deterministic channel modeling technique that can be used for CKM construction based on physical environment information. Ray tracing simulates electromagnetic wave propagation by tracking multiple paths between the Tx and Rx, accounting for reflection, diffraction, and scattering based on the environment’s geometry. Although theoretically precise, ray tracing suffers from prohibitive computational complexity and a strong dependence on accurate environmental information, such as geometry, dielectric properties, and material characteristics, thus limiting its practical implementation for large-scale and real-time applications. Recently, with the emergence of cellular networks and the Internet of Things, the massive number of connected devices can provide location-specific channel measurements. Combined with physical environment information, these measurement data can enable the efficient construction of CKMs via exploiting emerging data-driven approaches, achieving a balance between construction accuracy and computational efficiency. In the literature, representative CKM construction methods include interpolation, image processing and generative artificial intelligence (AI), as well as wireless radiance field (WRF) approaches. First, interpolation is widely used for constructing B2X CKMs, which estimates channel knowledge at unmeasured positions by using sparse measurement data from a limited number of reference points via techniques such as kernel regression, matrix completion, or Kriging [ 6 , 7 ] . These methods exploit spatial channel correlations and rely on the stationarity of the environment, thus making them unsuitable for high-dimensional X2X CKMs in complex environments. Next, image processing is another widely adopted approach for 2D B2X CKM construction based on sparse or low-resolution channel measurements, which treats the CKM as a gray-scale image, enabling the use of image processing and deep learning techniques like convolutional neural networks (CNNs) and super-resolution reconstruction [ 8 ] . Furthermore, generative AI techniques such as diffusion models have demonstrated strong capability for high-fidelity construction of 2D B2X CKM, which exploit the denoising mechanism to enable resolution enhancement and map completion (e.g., [ 9 , 10 ] ). More recently, inspired by the developments in computer graphics, wireless radiance field (WRF) frameworks have emerged as a new CKM construction method, by leveraging neural radiance fields (NeRF) and 3D Gaussian splatting (3DGS) to capture spatial and angular variations of wireless channels in continuous 3D space [ 11 , 12 ] . A WRF provides a continuous function mapping from 3D Tx/Rx position and propagation direction to channel knowledge, capturing environment-induced effects like reflection and scattering via proper models like neural networks or 3D Gaussian ellipsoids. On the one hand, NeRF-based WRF learns implicit volumetric representations by using multi-layer perceptrons (MLPs) to fit geometry and direction-dependent channel features. In contrast, 3DGS offers explicit parameterization through Gaussian ellipsoids, enabling efficient, high-resolution modeling of spatial variations and directional dependencies for complex 3D communication environments. These methods achieve high-fidelity CKM construction in 3D B2X scenarios by efficiently learning the joint spatial–angular channel mapping [ 11 , 12 ] , and can also be exploited for constructing the most comprehensive 6D X2X CKMs via the recently proposed bidirectional wireless Gaussian splatting (BiWGS) [ 13 ] . In light of the growing trend, this article provides a systematic overview on the recent advancements in CKM construction by particularly focusing on the three closely related approaches based on interpolation, image processing and generative AI, as well as the more recent WRF, respectively. For each approach, we first review representative methods and their underlying principles, and then discuss their application scenarios with different case studies. II Interpolation for CKM Construction Interpolation-based approaches have long been a conventional technique for constructing CKMs [ 3 ] , providing an efficient means of estimating channel knowledge at unmeasured locations from sparse measurements. In this case, CKMs are typically organized as geo-tagged databases, where channel knowledge parameters are indexed by geographic coordinates. The process of building such databases generally begins with data collection across the target area by using sparsely deployed sensors or devices. Interpolation is then performed to fill the measurement voids from sparsely distributed measurement. Within this framework, Kriging, kernel regression, and matrix completion are three widely used interpolation methods for CKM construction [ 6 ] . Kriging models spatial stationarity via variograms and produces