- The paper introduces a joint VOM-based CKM and monostatic ISAC protocol for accurate near-field channel estimation in 6G ELAA systems.
- It employs a Virtual Object Map to model static multipath components and dynamic sensing to extract transient scatterer paths, reducing pilot overhead.
- Empirical results demonstrate significantly lower NMSE and enhanced downlink rates compared to traditional methods.
Environment-Aware Near-Field Channel Estimation via CKM and ISAC
Problem Context and Motivation
The drive toward 6G wireless communication centers increasingly on extremely large-scale antenna arrays (ELAAs) operating in the near-field regime, which can, in principle, provide non-negligible spatial multiplexing gains and highly focused beamforming. However, the primary practical challenge remains the efficient and accurate estimation of high-dimensional near-field channels, especially under feedback- and resource-constrained scenarios. Traditional channel estimation protocols, heavily reliant on pilot-based acquisition, prove insufficient when facing the combinatorially large channel space induced by ELAA deployments. This work addresses these challenges by proposing a new framework that systematically leverages environment-aware priors—specifically, Channel Knowledge Maps (CKMs)—in concert with monostatic Integrated Sensing and Communication (ISAC) procedures. The framework is grounded on a Virtual Object Map (VOM) abstraction for static propagation elements and real-time dynamical information sensing for transient scatterers.
Technical Framework
The work is situated in a near-field ISAC system model: a base station (BS) equipped with an ELAA communicates with a single-antenna user equipment (UE) in a radiative near-field, quasi-static environment. The model expresses both communication and sensing channels as superpositions of multipath components (MPCs), each corresponding to a BS-visible primary interaction location with environment objects, described by three classes:
- Type-1 Environmental Objects (EOs): Compact, static scatterers (e.g., poles, signs).
- Type-2 EOs: Smooth, extended reflectors (e.g., building facades, ground).
- Sensing Targets (STs): Dynamic or transient environmental entities (e.g., vehicles, moving persons).
The channel models incorporate environment awareness by parametrizing the dominant static MPCs (due to static EOs) and the dynamic ones (due to STs). Critically, static MPCs are amenable to prior site-specific learning, while dynamic ones can only be sensed in real time.
Virtual Object Map (VOM) as CKM
The VOM is constructed offline using a virtual object library V—a catalogue of static BS-visible interaction points distilled via ray-tracing and clustering over the deployment site. For any UE or BS location, a VOM mapping produces (i) an index set of dominant static interaction points, ranked according to long-term channel contribution statistics, and (ii) the corresponding near-field array responses. This enables explicit parametric modeling of the static part of the channel during estimation.
Sensing-Assisted Protocol
Each channel coherence block is split into a pilot-assisted training phase and a data transmission phase. The BS transmits pilot sequences while simultaneously collecting monostatic echo measurements; the UE feeds back a quantized observation of received pilots using a codebook-based limited feedback strategy.
- Static Component: Handled using the VOM prior via the dominant static interaction points for the given UE and BS locations.
- Dynamic Component: Estimated by extracting a low-dimensional subspace from the clutter-suppressed monostatic echo (suppressing static VOM contributions), allowing for adaption to possibly extended and dynamic STs.
Channel Estimation Algorithm
The aggregate downlink channel is reconstructed as a weighted sum over the VOM-derived static responses and the sensed dynamic subspace basis. The coefficients for both components are jointly estimated by regularized least squares (ridge regression) from quantized pilot feedback and the extracted dynamic subspace basis. This structured approach drastically reduces estimation overhead compared to direct, unstructured channel estimation, as the search space is adapted to physical channel geometry and real-time environmental conditions.
Empirical Results and Key Numerical Findings
Comprehensive simulations on a canonical 64-antenna ELAA system at 2.4 GHz were performed, with both static EOs and dynamic STs distributed in a physically motivated fashion.
Channel Estimation Accuracy: The proposed joint VOM- and sensing-aided scheme achieves consistently superior normalized mean squared error (NMSE) in channel estimation over the full pilot-length regime. The gain over both the VOM-only and baseline (no environment awareness) methods is most significant in the short-pilot regime, highlighting the efficacy of physical priors in pilot-constrained settings.
Achievable Rate: The framework substantially improves the achievable downlink rate (under MRT beamforming), closely approaching the perfect-CSI bound for moderate and larger pilot budgets. Even static environment priors alone (VOM-only baseline) yield performance gains, indicating the centrality of site-specific CKMs in ELAA near-field operation.
Claim: The work asserts that joint exploitation of VOM-based static priors and real-time sensing of dynamic paths yields substantial gains in both channel estimation accuracy and system spectral efficiency over conventional methods that ignore such priors.
Implications and Future Research Directions
The results demonstrate that explicit incorporation of CKM/virtual-object environment priors and monostatic ISAC into channel estimation protocols can significantly alleviate training overhead and enhance reliability for 6G ELAA deployments. The VOM abstraction offers a scalable, physically interpretable, and hardware-compatible knowledge representation, supporting online adaptation as the underlying environment evolves.
Theoretical Implications: The separation of the static and dynamic channel estimation tasks, enabled by environment-aware priors and sensing, suggests new avenues in pilot design, subspace tracking, and channel feedback for high-dimensional non-stationary environments.
Practical Implications: Initial construction of robust VOMs is an offline process, but future work should consider online adaptive VOM updating in non-stationary or changing environments and proactive ISAC-based map enrichment. The proposed protocol is compatible with emerging feedback architectures and can extend naturally to multi-user scenarios, heterogeneous array geometries, and higher mobility regimes.
Conclusion
This paper introduces a rigorously designed environment-aware near-field channel estimation protocol for ELAA-equipped ISAC systems, combining CKM-based static priors through the VOM with monostatic sensing for dynamic paths. Simulation results decisively validate the superiority of this joint framework over both conventional and VOM-only designs, in channel estimation fidelity and end-to-end spectral efficiency. Future research should address online VOM construction, transferability across array topologies, and ISAC-based self-supervised refinement of environment priors.
