根據文獻記載，無線通訊系統可採用具有循環字首和循環字尾的特殊訓練序列架構，以實現聯合精細時間同步與通道估測；此一作法需要設定用於找出第一條通道路徑的搜尋範圍與門檻值。然而，在實際應用中，這些參數設定無法確保一定適當，因此對應的效能表現可能會明顯衰退。基於相同的訓練序列，本論文提出一種使用深度學習的聯合精細時間同步與通道估測方法，其中一個深度神經網路模型先在離線階段以大量的資料進行訓練，再運用至線上階段。所提出之方法不需要前述的參數設定，但具有相當出色的效能表現；模擬結果顯示，在使用3GPP短訓練序列的情況下，此方法在低訊號雜訊比之Extended Pedestrian A、Vehicular A與Typical Urban通道環境中，較之傳統作法，可分別改善 1.8%、 9.4% 和 37.2% 的正確時間同步機率；在使用3GPP長訓練序列的情況下，所提出之方法仍比傳統作法具有較佳的效能。隨著時間同步的改善，所提出之方法亦可達到更好的通道估測效能。

It is shown in the literature that joint fine time synchronization and channel estimation can be achieved for wireless communication systems by using a specific training sequence structure with both cyclic prefixing and cyclic postfixing. This method requires to set a search range and a threshold for determining the first tap of the channel impulse response. However, it is not possible to ensure appropriate settings for such parameters in practical applications, and the corresponding performance would be degraded significantly. In this thesis, a joint fine time synchronization and channel estimation scheme using deep learning (DL) is proposed based on the same training sequence structure. A deep neural network model is first trained offline by using a large number of data, and then the well-trained model is applied to the online stage. The proposed scheme is able to provide excellent performance without the above-mentioned parameter settings. Simulation results based on a 3rd Generation Partnership Project (3GPP) short training sequence demonstrate that the proposed DL-based method has nearly 1.8%, 9.4%, and 37.2% improvements in the probability of correct time synchronization over the conventional scheme at low signal-to-noise ratio situations under the Extended Pedestrian A, Vehicular A, and Typical Urban channel models, respectively. When a 3GPP long training sequence is used, the proposed method still achieves better time synchronization performance than the conventional scheme. With improved time synchronization, better channel estimation performance is achieved accordingly for the proposed DL-based scheme.

Abstract i
Contents ii
List of Figures iii
List of Tables v
I. Introduction 1
II. System Model 4
III. Review of Conventional Scheme for Joint Fine Time Synchronization and Channel Estimation 7
A. Coarse Time and Frequency Estimation 7
B. The Cross-Correlation Function Output 8
C. Joint Fine Time Synchronization and Channel Estimation Scheme 8
D. Issues Associated with the Conventional Scheme 10
IV. A Deep Learning Based Scheme for Joint Fine Time Synchronization and Channel Estimation 11
A. Deep Learning Basics 11
B. Joint Fine Time Synchronization and Channel Estimation Using Deep Learning 13
V. Simulation Results 16
VI. Conclusion 25
References 26

[1] C.-L. Wang and H.-C. Wang, “A low-complexity joint time synchronization and channel estimation scheme for orthogonal frequency division multiplexing systems,” in Proc. 2006 IEEE Int. Conf. Commun. (ICC 2006), Istanbul, Turkey, Jun. 2006, pp. 5670–5675.
[2] C.-L. Wang and H.-C. Wang, “On joint fine time adjustment and channel estimation for OFDM systems,” IEEE Trans. Wireless Commun., vol. 8, no. 10, pp. 4940–4944, Oct. 2009.
[3] Evolved Universal Terrestrial Radio Access (E-UTRA); Physical Channels and Modulation, 3GPP TS 36.211 V14.2.0, Apr. 2017.
[4] Physical Channels and Modulation, 3GPP TS 38.211 V15.2.0, Jul. 2018.
[5] T. O’Shea and J. Hoydis, “An introduction to deep learning for the physical layer,” IEEE Trans. Cogn. Commun. Netw., vol. 3, no. 4, pp. 563–575, Dec. 2017.
[6] T. Wang, C.-K. Wen, H. Wang, F. Gao, T. Jiang, and S. Jin, “Deep learning for wireless physical layer: opportunities and challenges,” China Commun., vol. 14, no. 11, pp. 92–111, Nov. 2017.
[7] O. Simeone, “A very brief introduction to machine learning with applications to communication systems,” IEEE Trans. Cogn. Commun. Netw., vol. 4, no. 4, pp. 648–664, Dec. 2018.
[8] M. Kulin, T. Kazaz, I. Moerman, and E. de Poorter, “A survey on machine learning-based performance improvement of wireless networks: PHY, MAC and network layer,” arXiv:2001.04561, Jan. 2020.
[9] H. Ye, G. Y. Li, and B. Juang, “Power of deep learning for channel estimation and signal detection in OFDM systems,” IEEE Commun. Lett., vol. 7, no. 1, pp. 114–117, Feb. 2018.
[10] E. Balevi and J. G. Andrews, “One-bit OFDM receivers via deep learning,” IEEE Trans. Commun., vol. 67, no. 6, pp. 4326–4336, Jun. 2019.
[11] K. Hornik, M. Stinchcombe, and H. White, “Multilayer feedforward networks are universal approximators,” Neural Netw., vol. 2, no. 5, pp. 359–366, Jul. 1989.
[12] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. Cambridge, MA, USA: MIT Press, 2016.
[13] D. P. Kingma and J. Ba, “Adam: a method for stochastic optimization,” in Proc. Int. Conf. Learn. Represent. (ICLR), San Diego, USA, May 2015, pp. 1–15.
[14] A. Chakrapani, “On the design details of SS/PBCH, signal generation and PRACH in 5G-NR,” IEEE Access, vol. 8, pp. 136617–136637, 2020.
[15] Evolved Universal Terrestrial Radio Access (E-UTRA); Base Station (BS) Radio Transmission and Reception, 3GPP TS 36.104 V14.3.0, Apr. 2017.