aiqbench analysis
This paper applies retrieval-augmented generation to wireless channel prediction: a Transformer draws on a store of historical CSI sequences from multiple scenes instead of storing all knowledge in its weights. The reported gains (over 10% across five metrics, including a 15.42% NMSE reduction) come from a single multi-scene dataset, so generalization to other deployments and hardware remains untested. For ML readers, the transferable idea is retrieval-based conditioning for time-series forecasting in physical systems, with limited added parameters.
RAG-CSI: Global Local Retrieval via Joint State and Dynamics Matching for CSI Prediction
arXiv:2610.36881v1 Announce Type: cross Abstract: Channel state information (CSI) prediction forecasts future channel states from historical observations, supporting transceiver design and resource allocation with reduced measurement overhead. However, existing deep learning predictors mainly encode channel knowledge in model parameters, which can limit their ability to represent diverse patterns of channel dynamics across scenes. In this paper, we propose RAG-CSI, a retrieval-enhanced Transformer framework that augments prediction with relevant historical channel knowledge. Specifically, we construct a global--local CSI knowledge store containing representative observations across scenes and historical observations from the current serving scene. We further develop joint state--dynamics retrieval (JSDR), which jointly characterizes channel states, phase dynamics, and amplitude dynamics to retrieve relevant CSI sequences. A Transformer predictor then uses cross-attention to fuse the retrieved knowledge with the current CSI history and predict future CSI with limited additional model parameters. Experiments on a multi-scene CSI dataset show overall improvements exceeding 10\% across all five evaluation metrics over the best baseline for each metric, including a 15.42\% reduction in normalized mean squared error.
Abstract:Channel state information (CSI) prediction forecasts future channel states from historical observations, supporting transceiver design and resource allocation with reduced measurement overhead. However, existing deep learning predictors mainly encode channel knowledge in model parameters, which can limit their ability to represent diverse patterns of channel dynamics across scenes. In this paper, we propose RAG-CSI, a retrieval-enhanced Transformer framework that augments prediction with relevant historical channel knowledge. Specifically, we construct a global--local CSI knowledge store containing representative observations across scenes and historical observations from the current serving scene. We further develop joint state--dynamics retrieval (JSDR), which jointly characterizes channel states, phase dynamics, and amplitude dynamics to retrieve relevant CSI sequences. A Transformer predictor then uses cross-attention to fuse the retrieved knowledge with the current CSI history and predict future CSI with limited additional model parameters. Experiments on a multi-scene CSI dataset show overall improvements exceeding 10\% across all five evaluation metrics over the best baseline for each metric, including a 15.42\% reduction in normalized mean squared error.
Signal Processing (eess.SP); Machine Learning (cs.LG)
arXiv:2610.36881 [eess.SP]
(or arXiv:2610.36881v1 [eess.SP] for this version)
https://doi.org/10.48550/arXiv.2610.36881 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 263 words
Related coverage
Predicting the Benefit of Retrieval Augmentation in Open-Domain Question Answering
arXiv · Aug 20, 2026
Predict, Then Retrieve: Cross-Instance Future-State Retrieval from Video Prefixes
arXiv · Aug 6, 2026
Benchmarking Retrieval-Augmented Generation Strategies for Large Language Model-Based Travel Mode Choice Prediction
arXiv · Aug 25, 2026
Back to the Future: Predicting Long-Horizon Dynamics with Retrospective Correction for Disturbance-Aware Retrieval
arXiv · Nov 3, 2026