深度學習(deep learning；DL)是一強而有力的基頻通訊系統設計技術。在本論文中，我們提出兩個DL方案來處理正交分頻多工(orthogonal frequency-division multiplexing；OFDM)系統的時間同步問題；第一個方案乃依據深度神經網絡(deep neural network；DNN)與一個包含核心序列以及循環字首和循環字尾的特定訓練序列架構而開發，適用於多輸入多輸出情境下的聯合精細時間同步與通道估測；此一DL方案具有優異的時間同步和通道估測效能，但卻無需如傳統的非DL作法使用統計通道資訊或設定其他的參數。第二個方案是依據兩個DNNs而開發，適用於單輸入多輸出情境下之5G NR傳輸中的前置序列偵測與定時提前(timing advance；TA)估測；一個DNN用於偵測前置序列是否存在，如果存在，另一個DNN將產生對應的TA估測值；此一雙步驟的DL方案能夠提供出色的前置序列偵測與TA估測結果，但卻無需如傳統的非 DL作法設置多個門檻值以降低干擾。大量的電腦模擬結果顯示，所提出的兩個DL方案都比傳統相關作法具有明顯更好的效能。

Deep learning (DL) is a powerful technique for baseband designs in communication systems. In this thesis, two DL-based schemes are proposed to deal with time synchronization problems in orthogonal frequency-division multiplexing (OFDM) systems. The first one is developed based on a deep neural network (DNN) and a specific training sequence structure, consisting of a kernel attached by both cyclic-prefix and cyclic-postfix extensions, for joint fine time synchronization and channel estimation in a multiple-input multiple-output scenario. The DL-based scheme works excellently for both time synchronization and channel estimation without statistical channel information or other parameter settings as required by the conventional (non-DL-based) method using the same training sequence structure. The second DL-based scheme is developed based on two DNNs for preamble detection and timing advance (TA) estimation in a single-input multiple-output scenario for 5G New Radio transmission. One DNN is used for detecting preamble existence; if a preamble is identified, the corresponding TA estimate is then generated by the other DNN. This two-step DL-based scheme is able to provide excellent results for preamble detection and TA estimation without threshold settings for interference reduction as required by the conventional (non-DL-based) method. Extensive simulation results demonstrate that both the proposed DL-based schemes offer much better performance than the conventional methods.

[1] S. B. Weinstein and P. M. Ebert, "Data transmission by frequency-division multiplexing using the discrete Fourier transform," IEEE Trans. Commun., vol. 19, no. 10, pp. 628–634, Oct. 1971.
[2] LTE; Evolved Universal Terrestrial Radio Access (E-UTRA); Physical Channels and Modulation, 3GPP TS 36.211 V14.2.0, Apr. 2017.
[3] 5G; NR; Physical Channels and Modulation, 3GPP TS 38.211V16.2.0, Jul. 2020.
[4] Radio Broadcasting Systems; Digital Audio Broadcasting (DAB) to Mobile, Portable and Fixed Receivers, ETSI EN 300 401 V2.1.1, Jan. 2017.
[5] Digital Video Broadcasting (DVB); Framing Structure, Channel Coding, and Modulation for Digital Terrestrial Television, ETSI EN 300 744 V1.6.2, Oct. 2015. [6] H. G. Myung, J. Lim, and D. J. Goodman, "Single carrier FDMA for uplink wireless transmission," IEEE Veh. Technol. Mag., vol. 1, no. 3, pp. 30–38, Sept. 2006.
[7] C. R. N. Athaudage, “BER sensitivity of OFDM systems to time synchronization error,” in Proc. 8th IEEE Int. Commun. Syst. (ICCS), Singapore, Nov. 2002, pp. 42–46.
[8] Y. Mostofi and D. Cox, “Mathematical analysis of the impact of timing synchronization errors on the performance of an OFDM system,” IEEE Trans. Commun., vol. 54, no. 2, pp. 226–230, Feb. 2006. [9] T. M. Schmidl and D. C. Cox, "Robust frequency and timing synchronization for OFDM," IEEE Trans. Commum., vol. 45, no. 12, pp. 1613–1621, Dec. 1997.
[10] H. S. Park, H. S. Cheon, C. G. Kang and D. S. Hong, “A simple preamble for OFDM timing offset estimation,” in Proc. IEEE 56th Veh. Technol. Conf. - Fall (VTC -Fall), vol. 2, Vancouver, BC, Canada, Sep. 2002, pp. 729–732.
[11] J.-J. van de Beek, M. Sandel, and P. O. Borjesson, “ML estimation of time and frequency offset in OFDM systems,” IEEE Trans. Signal Process., vol. 45, no. 7, pp. 1800–1805, Jul. 1997.
[12] J.-C. Lin, “Maximum-likelihood frame timing instant and frequency offset estimation for OFDM communication over a fast Rayleigh-fading channel,” IEEE Trans. Veh. Technol., vol. 52, no. 4, pp. 1049–1062, Jul. 2003.
[13] J. Lei and T.-S. Ng, “Pilot-tone-based maximum likelihood estimator for carrier frequency offset in OFDM systems,” in Proc. IEEE Int. Conf. Commun. (ICC), vol. 3, Anchorage, AK, USA, May. 2003, pp. 2046–2050.
[14] W. Lei, J.-H. Lu, and J. Gu, “A new pilot assisted frequency synchronization for wireless OFDM systems,” in Proc. IEEE Int. Conf. Acoust., Speech, Signal Process. (ICASSP), vol. 4, Hong Kong, China, Apr. 2003, pp. 700–703.
[15] LTE; Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall Description; Stage 2, 3GPP TS 36.300 V14.2.0 Apr. 2017
[16] 5G; NR; Overall Description; Stage-2, 3GPP TS 38.300 V15.8.0 Jan. 2020. [17] W. Guo, S. Hu, and S. Li, “Cyclic prefix-based time and frequency synchronization for TD-LTE downlink transmission,” in Proc. Sixth Int. Conf. Wireless Commun.Signal Process. (WCSP), Oct. 2014, pp. 1–5. [18] D. Wang, Z. Mei, H. Zhang, and H. Li, “A novel PSS timing synchronization algorithm for cell search in 5G NR system,” IEEE Access, vol. 9, pp. 5870–5880, Jun. 2021. [19] X. Yang and A. O. Fapojuwo, “Enhanced preamble detection for PRACH in LTE,” in Proc. IEEE Wireless Commun. Netw. Conf. (WCNC), Apr. 2013, pp. 3306–3311.
[20] 5G; NR; Physical Layer Procedures for Control, 3GPP TS 38.213 V15.6.0, Jul. 2019. [21] C.-L. Wang and H.-C. Wang, “Optimized joint fine timing synchronization and channel estimation for MIMO systems,” IEEE Trans. Commun., vol. 59, no. 4, pp. 1089–1098, Feb. 2011. [22] D. Magrin, C. Pielli, C. Stefanovic, and M. Zorzi, “Enabling LTE RACH collision multiplicity detection via machine learning,” in Proc. Int. Symp. Model. Optim. Mob. Ad Hoc Wireless Netw. (WiOpt), Jun. 2019, pp. 1–8.
[23] R. L. Frank and S. A. Zadoff, “Phase shift pulse codes with good periodic correlation properties,” IEEE Trans. Inf. Theory, vol. 8, no. 6, pp. 381–382, Oct. 1962.
[24] C. Chu, “Polyphase codes with good periodic correlation properties,” IEEE Trans. Inform. Theory, vol. 18, no. 4, pp. 531–532, Jul. 1972. [25] H. Ye, G. Y. Li and B. Juang, "Power of deep learning for channel estimation and signal detection in OFDM systems," IEEE Wireless Commun. Lett, vol. 7, no. 1, pp. 114–117, Feb. 2018.
[26] N. Farsad, M. Rao, and A. Goldsmith, “Deep learning for joint sourcechannel coding of text,” in Proc. IEEE Int. Conf. Acoust., Speech, Signal Process. (ICASSP), Apr. 2018, pp. 2326–2330. [27] S. Dörner, S. Cammerer, J. Hoydis and S. t. Brink, "Deep learning based communication over the air," IEEE J. Sel. Topics Signal Process., vol. 12, no. 1, pp. 132–143, Feb. 2018. [28] A. Alkhateeb, S. Alex, P. Varkey, Y. Li, Q. Qu, and D. Tujkovic, "Deep learning coordinated beamforming for highly-mobile millimeter wave systems," IEEE Access, vol. 6, pp. 37328–37348, Jun. 2018. [29] C.-C, Hsieh, “Joint fine time synchronization and channel estimation scheme using deep learning for wireless communication systems,” M.S. thesis, Inst. Commun. Eng., National Tsing Hua Univ., Hsinchu, Taiwan, Dec. 2021.
[30] K. Hornik, M. Stinchcombe, and H. White, “Multilayer feedforward networks are universal approximators,” Neural Netw., vol. 2, no. 5, pp. 359–366, Jul. 1989.
[31] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. Cambridge, MA, USA: MIT Press, 2016. [32] Charniak, E., Introduction to Deep Learning. Cambridge, MA, USA: The MIT Press, 2019 [33] A. L. Caterini and D. E. Chang, Deep Neural Networks in a Mathematical Framework. New York, NY, USA: Springer, 2018.
[34] L. Liu, H. Jiang, P. He, W. Chen, X. Liu, J. Gao, & J. Han, “On the variance of the adaptive learning rate and beyond, ” in Proc. Int. Conf. Learn. Represent. (ICLR), Addis Ababa, Ethiopia, Apr. 2020, pp. 1–13.
[35] 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.
[36] M. R. Zhang, J. Lucas, G. Hinton and J. Ba, “Lookahead optimizer: k steps forward, 1 step back,” arXiv:1907.08610, Dec. 2019. [37] S. Kabir, S. Sakib, M. A. Hossain, S. Islam, and M. I. Hossain, “A convolutional neural network based model with improved activation function and optimizer for effective intrusion detection and classification,” in Proc. Int. Conf. Adv. Comput. Innov. Technol. Eng. (ICACITE), Greater Noida, India, Mar. 2021, pp. 373–378. [38] G. Chen, Y. Zhang and J. Xing, "A method for recognizing feature regions of object model based on instance segmentation," in Proc. IEEE 4th Adv. Inf. Manag., Commun., Electron. Autom. Contr. Conf. (IMCEC), Chongqing, China, Jun. 2021, pp. 1114–1118.
[39] 5G; Study on Channel Model for Frequencies from 0.5 to 100 GHz, 3GPP TR 38.901 V14.0.0, May. 2017. [40] I. Bilbao and J. Bilbao, “Overfitting problem and the over-training in the era of data: Particularly for artificial neural networks,” in Proc. Eighth Int. Conf. Intell. Comput. Inf. Syst. (ICICIS), Dec. 2017, pp. 173–177. [41] G. E. Hinton, N. Srivastava, A. Krizhevsky, I. Sutskever, and R. R. Salakhutdinov, "Improving neural networks by preventing co-adaptation of feature detectors," arXiv:1207.0580, Jul. 2012. [42] J. Kukačka, V. Golkov, and D. Cremers, “Regularization for deep learning: A taxonomy,” arXiv: 1710.10686, Oct. 2017.
[43] N. Modina, R. Ferrari, and M. Magarini, “A machine learning-based design of PRACH receiver in 5G,” Procedia Comput. Sci., vol. 151, pp. 1100–1107, Apr. 2019. [44] T. A. Pham and B. T. Le, “A proposed preamble detection algorithm for 5G-PRACH,” in Proc. Int. Conf. Adv. Technol. Commun. (ATC), Oct. 2019, pp. 210–214.
[45] 5G; NR; Medium Access Control (MAC) Protocol Specification, 3GPP TS 38.321 V16.3.0, Jan. 2021. [46] S. Sesia, M. Baker, and I. Toufik, LTE - The UMTS Long Term Evolution: From Theory to Practice, 2nd ed. Hoboken, NJ, USA: John Wiley & Sons, 2011.
[47] 5G; NR; Base Station (BS) conformance testing Part 1: Conducted conformance testing, 3GPP TS 38.141-1 V16.6.0, Jan. 2021.