近年來，隨著通訊技術的日益發展，第五代行動通訊技術（5G）也逐漸普及。這樣的背景下，解碼技術仍不斷追求高可靠度和低延遲的需求。在5G規格中，極化碼被作為錯誤更正碼的編碼方案之一，而在極化碼的解碼方法中，順續消去法（SC）和改良版的順續消去列表法（SCL）被廣泛應用。
然而，順續消去法解碼時需要按照順續進行，且下一個位元解碼時需要使用前一個位元解碼的資訊。這樣一來，如果前面的位元解碼錯誤，就會導致錯誤資訊不斷傳遞。為了改善這個問題，順續消去法翻轉解碼（SC-Flip）應運而生。它在初次解碼失敗後，將解碼至葉節點的訊息位元的絕對值對數似然比（|LLR|）作為指標，透過數次重新進行SC解碼並在過程中翻轉一個錯誤機率高的位元，以避免錯誤傳遞。然而，我們希望使用基於LSTM神經網路的判斷方式，來用不同的方法來改善錯誤的傳遞。
本篇論文的主要目標是改善順續消去法翻轉解碼中翻轉判斷的標準，並使用深度學習的方式提供輔助。研究中利用長短期記憶模型（LSTM）進行訓練，同時對其它篇論文中的訓練方法進行改良。最終，將這種改進的解碼方法應用於更高可靠度的順續消去列表法(SCL)中，並結合循環冗餘校驗（CRC）因子，我們稱LSTM-CA-SCLF。研究中根據不同的訊雜比進行多次翻轉的探討，以降低整體區塊錯誤率（BLER），同時盡量減少平均解碼回數（ADR）的影響。

In recent years, the rapid advancement of communication technology has led to the widespread adoption of the fifth-generation mobile communication technology(5G). Within this context, there is a growing demand for decoding techniques that offer high reliability and low latency. Polar codes have emerged as one of the coding schemes for error correction in the 5G specification, with successive cancellation (SC) and the enhanced successive cancellation list (SCL) being widely employed for decoding polar codes.

However, successive cancellation decoding is performed in a sequential order, and the decoding of the next bit requires information from the decoding of the previous bit. As a result, if there is an error in the decoding of a previous bit, it will lead to the propagation of errors. To address this problem, successive cancellation with flipping decoding (SC-Flip) was introduced. After the initial decoding failure, SC-Flip uses the absolute value of the log-likelihood ratio (|LLR|) of the decoded bit at the leaf node as an indicator. It performs multiple iterations of SC decoding and flips a bit with a high error probability during the process to prevent error propagation. However, we aim to improve the error propagation using a judgment method based on the LSTM neural network, employing different approaches.

The primary objective of this paper is to enhance the criteria used for flipping judgment in successive cancellation with flipping decoding and provide auxiliary support through deep learning. In this study, we train a long short-term memory (LSTM) model and improve upon existing training methods proposed in related literature. Ultimately, we apply this enhanced decoding approach to the more reliable successive cancellation list (SCL), incorporating cyclic redundancy check (CRC) factors, referred to as LSTM-CA-SCLF. Our research investigates multiple iterations of flipping based on various signal-to-noise ratios (SNR) to minimize the overall block error rate (BLER) while mitigating the impact on the average decoding round (ADR).

[1] E. Arikan, "Channel polarization: A method for constructing
capacity-achieving codes," IEEE International Symposium on
Information Theory, pp. 1173-1177, 2008.

[2] E. Arikan, "Channel Polarization: A Method for Constructing
Capacity-Achieving Codes for Symmetric Binary-Input
Memoryless Channels, " IEEE Transactions on Information
Theory, vol. 55, no. 7, pp. 3051-3073, 2009.

[3] O. Afisiadis, A. Balatsoukas-Stimming, and A. Burg, "A Low-Complexity Improved Successive Cancellation Decoder for Polar Codes," 2014 48th Asilomar Conference on Signals, Systems and Computers, pp. 2116-2120, 2014.

[4] A. Balatsoukas-Stimming, M. B. Parizi, and A. Burg, "LLRBased
Successive Cancellation List Decoding of Polar Codes,"
IEEE Transactions on Signal Processing, vol. 63, no. 19, pp.
5165-5179, 2015.

[5] Y. Yongrun, P. Zhiwen, L. Nan and Y. Xiaohu, "Successive Cancellation List Bit-flip Decoder for Polar Codes," 2018 10th International Conference on Wireless Communications and Signal Processing (WCSP), pp. 1-6, 2018.

[6] F. Cheng, A. Liu, Y. Zhang, and J. Ren, "Bit-Flip Algorithm for Successive Cancellation List Decoder of Polar Codes," in IEEE Access, vol. 7, pp. 58346-58352, 2019.

[7] K. Niu and K. Chen, "CRC-Aided Decoding of Polar Codes," IEEE Communications Letters, vol. 16, no. 10, pp.
1668-1671, 2012.

[8] 呂育瑋，“高效快速可適性極化碼解碼搭配分段CRC 順續消除列表法，”國立台灣科技大學電子工程所，2020。

[9] X. Wang, H. Zhang, R. Li, L. Huang, S. Dai, Y. Huangfu, and J. Wang, "Learning to Flip Successive Cancellation Decoding of Polar Codes with LSTM Networks," IEEE 30th Annual Inter. Sym. on Per., Indoor and Mobile Radio Commun. (PIMRC), pp.1-5, Nov. 2019.

[10] E. Arikan, "A Machine Learning Based Multi-flips Successive Cancellation Decoding Scheme of Polar Codes," 2020 IEEE 91st Vehicular Technology Conference (VTC2020-Spring), pp. 1-5, Jun. 2020.

[11] F. Carpi, C. Hager, M. Martalo, R. Raheli and H. D. Pfister, "Reinforcement Learning for Channel Coding: Learned Bit-Flipping Decoding," 2019 57th Annual Allerton Conference on Communication, pp. 922-929, 2019.

[12] R. DiPietro and GD. Hager, "Deep learning: RNNs and LSTM," Handbook of Medical Image Computing and Computer Assisted Intervention, pp. 503-519, 2019.