Delay-Doppler Sparsity-Constrained U-Net for OTFS Channel Estimation under Fractional Doppler
Authors
Riccardo Amato, Stefano Di Marco, Antonio Ferrara, Emiliano GentileFiles
Abstract
Fractional Doppler distributes the response of a propagation path across neighboring delay-Doppler cells, complicating sparse channel estimation in orthogonal time-frequency space systems. This study presents a residual U-Net that combines local Doppler neighborhoods with observation consistency to reconstruct an effective sampled channel. Its four input channels contain the real and imaginary components of a pilot-based initialization, a visibility mask, and a noise scale map. A piecewise differentiable shrinkage stage uses shared neighborhood energy to regulate the reconstructed support while preserving the phase of retained coefficients. Five estimator configurations are compared within a common evaluation specification. At a nominal signal-to-noise ratio of 20 dB, the reported normalized mean square error is −22.4 dB for the constrained configuration and −18.8 dB for the unconstrained U-Net, a separation of 3.6 dB. The corresponding support precision and recall are 0.91 and 0.89, compared with 0.82 and 0.86 for the unconstrained configuration. Ablation, offset, and path-density comparisons examine the relationship between neighborhood sharing and leakage reconstruction. The contribution concerns the integration of residual reconstruction, noise-conditioned neighborhood shrinkage, and an explicit observation penalty under integer-aligned delays. The reported coordinates describe the numerical comparison; the specified training allocation and device-level receiver validation remain separate implementation requirements. Practical assessment should examine pulse conventions, synchronization mismatch, fractional delay, and fixed-detector performance. These conditions define the scope of the present channel-domain interpretation.
References
Aldababsa, M., Özyurt, S., Kurt, G. K., & Kucur, O. (2024). A survey on orthogonal time frequency space modulation. IEEE Open Journal of the Communications Society, 5, 4483–4518. https://doi.org/10.1109/ojcoms.2024.3422801
Sheng, H.-T., & Wu, W.-R. (2024). Time-frequency domain channel estimation for OTFS systems. IEEE Transactions on Wireless Communications, 23(2), 937–948. https://doi.org/10.1109/twc.2023.3283578
Jesbin, F., & Chockalingam, A. (2026). Delay-Doppler pulse shaping in Zak-OTFS using Hermite basis functions. IEEE Transactions on Vehicular Technology, 75(10), 23804–23817. https://doi.org/10.1109/tvt.2026.3703829
Liu, C., Xiang, W., Long, H., & Jia, Y. (2024). Spectral-efficient reference signal design and iterative channel estimation for OTFS modulation. IEEE Transactions on Vehicular Technology, 73(10), 15286–15301. https://doi.org/10.1109/tvt.2024.3414468
Wang, X., Zheng, C., Hu, P., Yang, J., & Kang, C. G. (2025). A sparsity-agnostic SL0 channel estimation approach for OTFS systems. IEEE Communications Letters, 29(5), 1097–1101. https://doi.org/10.1109/lcomm.2025.3554379
Zegrar, S. E., Sümer, A. S., & Arslan, H. (2025). Fractional delay and fractional Doppler estimation and mitigation framework in OTFS systems. IEEE Transactions on Vehicular Technology, 74(2), 2884–2896. https://doi.org/10.1109/tvt.2024.3470839
Lee, H.-G., Kim, J., Zhao, J., & Joung, J. (2025). Single-tone-based channel estimation method for OTFS systems. IEEE Transactions on Communications, 73(3), 1638–1651. https://doi.org/10.1109/tcomm.2024.3455240
Li, W., Zhou, D., & Ye, Q. (2025). Channel computation based on multi-scale attention residual network. Journal of Information and Intelligence, 3(3), 275–287. https://doi.org/10.1016/j.jiixd.2025.03.001
Al-Imran, Nazim, M. S., Nguyen, H., & Jang, Y. M. (2025). Channel estimation of massive MIMO FSO communication system using deep attention residual U-Net. ICT Express, 11(2), 287–292. https://doi.org/10.1016/j.icte.2024.09.012
Wang, Q., Chen, X., Tao, Q., Qian, L. P., Kam, P.-Y., & Wu, Y. (2025). Model-driven channel estimation network for orthogonal time-frequency space systems. IEEE Transactions on Vehicular Technology, 74(7), 11552–11556. https://doi.org/10.1109/tvt.2025.3548639
He, Y., & Joseph, G. (2025). Bayesian algorithms for Kronecker-structured sparse vector recovery with application to IRS-MIMO channel estimation. IEEE Transactions on Signal Processing, 73, 142–157. https://doi.org/10.1109/tsp.2024.3416543
Luan, D., & Thompson, J. S. (2026). Achieving robust channel estimation neural networks by designed training data. IEEE Transactions on Cognitive Communications and Networking, 12, 2311–2324. https://doi.org/10.1109/tccn.2025.3592299
Zhang, H., Huang, X., & Zhang, J. A. (2024). Low-overhead OTFS transmission with frequency or time domain channel estimation. IEEE Transactions on Vehicular Technology, 73(1), 799–811. https://doi.org/10.1109/tvt.2023.3305921
Wang, R., Ren, H., Pan, C., Weng, R., & Wang, J. (2026). A cyclic shift embedded pilot-based channel estimation for multiuser MIMO-OTFS systems with fractional delay and Doppler. IEEE Transactions on Signal Processing, 74, 4466–4482. https://doi.org/10.1109/tsp.2026.3721735
He, X., Yuan, W., & Fan, P. (2024). On the pilot-aided channel estimation for windowed OTFS with data interference in rapidly time-varying channels. IEEE Transactions on Wireless Communications, 23(11), 16359–16374. https://doi.org/10.1109/twc.2024.3440586
Tang, M., Wang, H., Yuan, Z., & Yuan, J. (2024). A novel off-grid channel estimation with fast BCS using LSM prior for OTFS modulation. IEEE Transactions on Wireless Communications, 23(9), 12157–12171. https://doi.org/10.1109/twc.2024.3388449
Guo, Q., Jiang, H., Xiang, J., & Zhong, Y. (2024). A CS-BEM OTFS channel estimation approach for sparse continuous Doppler-spread channels. IEEE Wireless Communications Letters, 13(11), 2985–2989. https://doi.org/10.1109/lwc.2024.3432090
Lan, X., Zhang, X., Zhang, L., Ma, Z., Wang, Q., & Akan, O. B. (2026). A sparse Bayesian learning-based OTFS channel estimation for 6G wireless system. IEEE Wireless Communications Letters, 15, 805–809. https://doi.org/10.1109/lwc.2025.3642986
Meng, K., Yang, S., Wang, X.-Y., Bu, Y., Tang, Y., Zhang, J., & Hanzo, L. (2024). Joint sparsity pattern learning based channel estimation for massive MIMO-OTFS systems. IEEE Transactions on Vehicular Technology, 73(8), 12189–12194. https://doi.org/10.1109/tvt.2024.3375027
Mattu, S. R., & Chockalingam, A. (2024). Learning in time-frequency domain for fractional delay-Doppler channel estimation in OTFS. IEEE Wireless Communications Letters, 13(5), 1245–1249. https://doi.org/10.1109/lwc.2024.3367112
Gupta, C., & Yadav, S. S. (2025). Deep learning based channel estimation for UAVs: A modified U-Net approach. Advances in Electrical and Computer Engineering, 25(1), 61–70. https://doi.org/10.4316/aece.2025.01007
Wang, X., Zhang, H., Yang, Q., Wang, B., Gui, G., Yuen, C., & Sari, H. (2026). Model-based deep unfolding for efficient zero-padded OTFS channel estimation. IEEE Transactions on Vehicular Technology, 75(1), 756–768. https://doi.org/10.1109/tvt.2025.3592422
Lee, J. W., & Choi, K. W. (2026). Two-tier PEFT adapter switching for environment- and SNR-adaptive channel estimation. IEEE Open Journal of the Communications Society, 7, 6024–6046. https://doi.org/10.1109/ojcoms.2026.3697928
Das, A., Jesbin, F., & Chockalingam, A. (2026). A Gaussian-sinc pulse shaping filter for Zak-OTFS. IEEE Transactions on Vehicular Technology, 75(3), 4307–4322. https://doi.org/10.1109/tvt.2025.3609781
Zhou, M., Chen, F., Xia, M., Zhang, X., & Yu, H. (2025). Iterative channel estimation for multi-user OTFS uplink systems with superimposed full pilots. IEEE Transactions on Vehicular Technology, 74(3), 4485–4497. https://doi.org/10.1109/tvt.2024.3497176
Ying, D., & Ye, F. (2025). Deep learning supported path prediction and channel estimation for MIMO-OTFS system with high delay resolution. IEEE Transactions on Vehicular Technology, 74(3), 3584–3597. https://doi.org/10.1109/tvt.2024.3493921
Mehrotra, A., Srivastava, S., Shanmughanadha Reddy, N., Jagannatham, A., & Hanzo, L. (2025). Sparse channel estimation for MIMO OTFS/OTSM systems using finite-resolution ADCs. IEEE Transactions on Communications, 73(6), 3971–3987. https://doi.org/10.1109/tcomm.2024.3502682
Zhang, X., Liu, C., Yuan, W., Zhang, J. A., & Ng, D. W. K. (2024). Sparse prior-guided deep learning for OTFS channel estimation. IEEE Transactions on Vehicular Technology, 73(12), 19913–19918. https://doi.org/10.1109/tvt.2024.3450012
Li, X., Fan, P., Wang, Q., & Liu, Z. (2025). Grid evolution for doubly fractional channel estimation in OTFS systems. IEEE Transactions on Vehicular Technology, 74(2), 3486–3490. https://doi.org/10.1109/tvt.2024.3472695
Kong, L., Liu, X., Zhang, X., Xiong, J., Zhao, H., & Wei, J. (2025). Representation-based continual learning for channel estimation in dynamic wireless environments. IEEE Transactions on Wireless Communications, 24(8), 6382–6396. https://doi.org/10.1109/twc.2025.3552836
Ma, M., Dai, J., & Jiang, X.-Q. (2025). Fast burst-sparsity learning approach for massive MIMO-OTFS channel estimation. IEEE Wireless Communications Letters, 14(5), 1546–1550. https://doi.org/10.1109/lwc.2025.3548685
Ito, K., Takahashi, T., Furuta, K., Ibi, S., & Thadeu Freitas de Abreu, G. (2024). Joint channel and data estimation via parametric bilinear inference for OTFS demodulation. IEEE Open Journal of the Communications Society, 5, 6090–6105. https://doi.org/10.1109/ojcoms.2024.3463202
Yue, Y., Shi, J., Li, Z., Hu, J., & Tie, Z. (2024). Model-driven deep learning assisted detector for OTFS with channel estimation error. IEEE Communications Letters, 28(4), 842–846. https://doi.org/10.1109/lcomm.2024.3362970