DRIFT: Joint Channel Estimation and Prediction Towards Pilotless 6G Non-Terrestrial Networks
Bruno De Filippo 1 , Carla Amatetti 1 , Alessandro Vanelli-Coralli 1 · 2026
Read DRIFT: Joint Channel Estimation and Prediction Towards Pilotless 6G Non-Terrestrial Networks by Bruno De Filippo 1 , Carla Amatetti 1 , Alessandro Vanelli-Coralli 1 (2026) 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-2605-31065.txt.
Introduction
De Filippo and colleagues design channel prediction for satellite networks in 6G systems. Channel prediction limits pilot overhead, but AI predictors are often too complex for onboard use under power limits. Their DRIFT framework jointly estimates and predicts channels, sending pilots only in the first slot and using data processing after that. A lightweight architecture refines data estimates and predicts future responses with low cost and reduced error spread. Simulations show spectrum gains over pilot systems, with mismatch robustness and few operations for onboard use.
Catalog record
| CREATOR | Bruno De Filippo 1 , Carla Amatetti 1 , Alessandro Vanelli-Coralli 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25674 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.31065
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
DRIFT: Joint Channel Estimation and Prediction Towards Pilotless 6G Non-Terrestrial Networks Bruno De Filippo 1 , Carla Amatetti 1 , Alessandro Vanelli-Coralli 1 Abstract Non-terrestrial networks (NTNs) are expected to play a pivotal role in sixth-generation (6G) systems by enabling ubiquitous connectivity and massive communication. In this context, channel prediction emerges as a key technique to improve the spectrum utilization efficiency by limiting the pilot overhead. However, many proposed predictors based on artificial intelligence (AI) are characterized by high inference complexity, posing challenges to onboard implementation. In this paper, we address the challenge of designing accurate yet computationally efficient channel prediction techniques tailored to low Earth orbit (LEO) NTNs, where strict power constraints limit model complexity, to enable spectral efficiency gains. We propose an iterative joint channel estimation and prediction framework in the context of 6G NTNs that significantly reduces pilot overhead by transmitting pilots only in the initial slot and relying on data-driven processing for subsequent slots. We introduce Data-driven Refinement and Iterative Forecast for wireless channel Tracking (DRIFT), a lightweight architecture that refines data-aided channel estimates and predicts future channel frequency responses with low computational cost and reduced error propagation. Two predictor variants based on convolutional and long short-term memory layers are investigated. Simulation results in an end-to-end simulation of an uplink LEO NTN scenario show that the proposed approach achieves up to 12% spectral efficiency gain compared to conventional pilot-based systems, with robustness to training-test mismatches and consistent performance across different channel models. Moreover, DRIFT requires fewer than 200k multiply-accumulate operations, making it suitable for on-board satellite implementation under stringent power constraints. {bruno.defilippo, carla.amatetti2, alessandro.vanelli}@unibo.it I Introduction Non-terrestrial networks (NTNs), encompassing satellite, high-altitude platform, and unmanned aerial vehicle communications, are emerging as a cornerstone of future sixth-generation (6G) systems due to their ability to provide ubiquitous connectivity beyond the limits of terrestrial infrastructure [ 8 ] . At the same time, artificial intelligence (AI) is widely recognized as one of the fundamental technologies for the evolution from 5G to 6G, with new base stations being denoted in the 3rd generation partnership project (3GPP) as aNodeBs (aNBs) as a possible reference to the AI-native nature of 6G [ 1 ] . In particular, technical report 38.843 has identified three main use cases for the application of AI in the radio access network, among which is channel prediction [ 3 ] . This is particularly relevant in the context of low Earth orbit (LEO) NTNs, where the orbit dynamics introduce strong Doppler shifts and delays in a time-varying but deterministic manner. Indeed, several authors have proposed channel prediction solutions to exploit such dynamics with the objective of countering the so-called channel aging effect, e.g. , with a long short-term memory (LSTM)-based model in [ 4 ] and with a convolutional neural network (CNN) and an LSTM layer in [ 14 ] . A transformer-based model was proposed in [ 13 ] to predict a time series of channel frequency response (CFR) matrices based on historical data in multi-antenna LEO-based NTNs, achieving state-of-the-art normalized mean squared error (NMSE) figures while requiring more than 100M floating point operations (FLOPs) for inference. In [ 15 ] , the time series prediction task was carried out with a simpler LSTM-based model. An LSTM-based model was also used in [ 9 ] to predict the excess path loss in Q/V band NTN links. Finally, the authors in [ 6 ] incorporated the attention mechanism in a CNN+LSTM model taking into account not only historical channel data, but also the LEO satellite’s elevation angle. Differently from terrestrial networks, NTNs typically operate under strict power budget constraints, severely limiting the computational complexity of deep learning models. Thus, the design of a channel predictor tailored to NTNs is a tradeoff between prediction accuracy and computational complexity. Nonetheless, most of the works in the literature tend to emphasize to former over the latter. Furthermore, the impact of channel prediction inaccuracies on end-to-end metrics such as the effective spectral efficiency (SE) and throughput is rarely evaluated, with many works focusing on the NMSE instead. In our previous works [ 5 , 7 ] we aimed at covering such aspects by proposing lightweight channel predictors based on temporal convolutional networks (TCNs) and CNN-LSTM tailored to NTNs with the objective of reducing the pilot overhead and, thus, improving the throughput by up to \approx 8.33 %. In this paper, we build on top of such works by: • Presenting an iterative joint channel estimation and prediction framework where pilot-aided estimation is replaced with channel refinement and prediction; • Proposing the Data-driven Refinement and Iterative Forecast for wireless channel Tracking (DRIFT) model, which refines a data-aided (DA) channel estimates and predicts the upcoming CFR with extremely low complexity; • Evaluating the SE gains achieved with respect to traditional estimation-based systems in a simulated LEO-based NTN link, also considering mismatches between training and test data. II System model We consider an uplink transmission from a single user equipment (UE) to an aNB hosted on a LEO satellite, as in a 6G NTN scenario. The UE employs a cyclic prefix orthogonal frequency division multiplexing (CP-OFDM) waveform and is assumed to pre-compensate the deterministic Doppler and delay components induced by the satellite mobility through a global navigation satellite system (GNSS) receiver, so that only a residual frequency synchronization error affects the link. The transmitter chain includes low-density parity-check (LDPC) channel encoding, bit interleaving, M -ary quadrature amplitude modulation (QAM) mapping, and CP-OFDM multiplexing. Let N_{SC} denote the number of subcarriers allocated to the UE, and let N_{sym}^{slot} be the number of OFDM symbols per slot. The transmitted resource grid in slot t is represented as: \mathbf{X}_{t}\in\mathbb{C}^{N_{SC}\times N_{sym}^{slot}}. (1) We model the time-frequency-selective channel as a tapped-delay line (TDL) with a residual carrier frequency offset \xi\sim\mathcal{N}(0,\sigma_{f}^{2}) due to imperfect frequency synchronization, with \sigma_{f}^{2} being the variance of the carrier frequency offset. Under the assumption that \xi is small compared to the subcarrier spacing \Delta f , the received grid can be obtained as: \mathbf{Y}_{t}=\mathbf{H}_{t}\odot\mathbf{X}_{t}+\mathbf{W}_{t}, (2) where \mathbf{H}_{t}\in\mathbb{C}^{N_{SC}\times N_{sym}^{slot}} is the CFR matrix in slot t , and \odot denotes the Hadamard product. \mathbf{W}_{t}\in\mathbb{C}^{N_{SC}\times N_{sym}^{slot}} models additive white Gaussian noise (AWGN), i.e. , \left(\mathbf{W}_{t}\right)_{k,n}\sim\mathcal{N}(0,N_{0}/2) , where \left(\mathbf{W}_{t}\right)_{k,n} denotes the element at the k -th subcarrier and n -th OFDM symbol of \mathbf{W}_{t} and N_{0} is the receiver’s noise power spectral density. Fig. 1: Iterative joint channel estimation and prediction framework. The proposed receiver operates in an iterative prediction loop across N_{slot} consecutive slots. In the first slot, denoted by t=1 , pilot symbols are inserted in the resource grid to initialize the process as in a traditional estimation-based system. Specifically, we alternate quadrature phase shift keying (QPSK) pilot symbols with null symbols over the subcarriers at OFDM symbol indices \mathcal{I}_{\pi}=\{3,12\} (1-based indexing), therefore boosting the pilot power. All the remaining resource elements of the first slot carry data symbols. For the subsequent slots, 2\leq t\leq N_{slot} , no pilots are transmitted and the resource grid carries only data symbols. Therefore, the theoretical SE gain achieved through this framework can be evaluated as: \Gamma(N_{slot})=\frac{N_{slot}N_{sym}^{slot}-\lvert\mathcal{I}_{\pi}\rvert}{N_{slot}\left(N_{sym}^{slot}-\lvert\mathcal{I}_{\pi}\rvert\right)}-1. (3) It is worth noting that, for an infinite sequence of pilotless slots, such gain approaches \Gamma_{\infty}=N_{sym}^{slot}/(N_{sym}^{slot}-\lvert\mathcal{I}_{\pi}\rvert)-1\approx 16.67 %. Nonetheless, the formulation in (