Fronthaul-Efficient Distributed Cooperative 3D Positioning with Quantized Latent CSI Embeddings

TL;DR

This work tackles the fronthaul bottleneck in cooperative 3D positioning by introducing an edge-cloud architecture that compresses locally estimated CSI into fixed-length latent embeddings transmitted from multiple BSs. The central unit fuses these quantized embeddings via a channel-masked attention mechanism and a frequency-evidence sequence model to jointly infer 3D user position without reconstructing raw CSI. A two-stage training pipeline—self-supervised edge embedding at the BSs followed by end-to-end joint training with STE through the quantization bottleneck—enables reliable performance under tight fronthaul budgets. In a 3.5 GHz urban ray-tracing scenario with six BSs and 20 MHz bandwidth, the approach achieves sub-meter mean positioning error (≈0.48 m) and 90th percentile error (≈0.83 m) while reducing fronthaul payload to about 6.25% of lossless CSI, approaching the performance of full CSI exchange. These results underscore the practicality of task-oriented CSI representations for scalable cooperative localization in dense urban environments.

Abstract

High-precision three-dimensional (3D) positioning in dense urban non-line-of-sight (NLOS) environments benefits significantly from cooperation among multiple distributed base stations (BSs). However, forwarding raw CSI from multiple BSs to a central unit (CU) incurs prohibitive fronthaul overhead, which limits scalable cooperative positioning in practice. This paper proposes a learning-based edge-cloud cooperative positioning framework under limited-capacity fronthaul constraints. In the proposed architecture, a neural network is deployed at each BS to compress the locally estimated CSI into a quantized representation subject to a fixed fronthaul payload. The quantized CSI is transmitted to the CU, which performs cooperative 3D positioning by jointly processing the compressed CSI received from multiple BSs. The proposed framework adopts a two-stage training strategy consisting of self-supervised local training at the BSs and end-to-end joint training for positioning at the CU. Simulation results based on a 3.5~GHz 5G NR compliant urban ray-tracing scenario with six BSs and 20~MHz bandwidth show that the proposed method achieves a mean 3D positioning error of 0.48~m and a 90th-percentile error of 0.83~m, while reducing the fronthaul payload to 6.25% of lossless CSI forwarding. The achieved performance is close to that of cooperative positioning with full CSI exchange.

Figures (14)

• Figure 1: Cloud-RAN architecture for cooperative 3D positioning in urban NLOS environments.
• Figure 2: The proposed ECC positioning framework. At the edges, each BS estimates the local CSI from uplink and transform it to quantized latent embedding. At the cloud, the CU infers 3D positioning of UE directly from the received bitstreams.
• Figure 3: The proposed embedding network (Res-SegNet).
• Figure 4: The proposed CMA mechanism.
• Figure 5: The proposed frequency evidence accumulation mechanism.