More Like this

1. Data-Driven List Polar Decoder for Symmetric and Asymmetric Input Distributions

   Aharoni, Ziv; Huleihel, Bashar; Pfister, Henry D; Permuter, Haim H (March 2024, ETH Zurich)

   Lapidoth, Amos; Moser, Stefan M (Ed.)

   This paper introduces extensions to data-driven polar decoders, enabling list decoding and accommodating asymmetric input distributions. These are crucial steps to develop data-driven codes that 1) achieve capacity and 2) are competitive in moderate block lengths. We commence by integrating list de- coding into the data-driven polar codes, which significantly alleviates the inherent error propagation issues associated with successive cancellation decoding. Secondly, we expand the applicability of these codes to channels with stationary, non-uniform input distributions by incorporating the Honda-Yamamoto scheme. Both modifications are computationally efficient and do not require an explicit channel model. Numerical results validate the efficacy of our contributions, which offer a robust and versatile coding mechanism for various channel conditions. 

   more » « less
2. Capacity-Achieving Polar-Based Codes With Sparsity Constraints on the Generator Matrices

   https://doi.org/10.1109/TCOMM.2023.3282603  

   Pang, James Chin-Jen; Mahdavifar, Hessam; Pradhan, S. Sandeep (September 2023, IEEE Transactions on Communications)

   In general, the generator matrix sparsity is a critical factor in determining the encoding complexity of a linear code. Further, certain applications, e.g., distributed crowdsourcing schemes utilizing linear codes, require most or even all the columns of the generator matrix to have some degree of sparsity. In this paper, we leverage polar codes and the well-established channel polarization to design capacity-achieving codes with a certain constraint on the weights of all the columns in the generator matrix (GM) while having a low-complexity decoding algorithm. We first show that given a binary-input memoryless symmetric (BMS) channel $$W$$ and a constant $$s \in (0, 1]$$ , there exists a polarization kernel such that the corresponding polar code is capacity-achieving with the rate of polarization $s/2$ , and the GM column weights being bounded from above by $$N^{s}$$ . To improve the sparsity versus error rate trade-off, we devise a column-splitting algorithm and two coding schemes for BEC and then for general BMS channels. The polar-based codes generated by the two schemes inherit several fundamental properties of polar codes with the original $$2 \times 2$$ kernel including the decay in error probability, decoding complexity, and the capacity-achieving property. Furthermore, they demonstrate the additional property that their GM column weights are bounded from above sublinearly in $$N$$ , while the original polar codes have some column weights that are linear in $$N$$ . In particular, for any BEC and $$\beta < 0.5$$ , the existence of a sequence of capacity-achieving polar-based codes where all the GM column weights are bounded from above by $$N^{\lambda} $$ with $$\lambda \approx 0.585$$ , and with the error probability bounded by $${\mathcal {O}}(2^{-N^{\beta }})$$ under a decoder with complexity $${\mathcal {O}}(N\log N)$$ , is shown. The existence of similar capacity-achieving polar-based codes with the same decoding complexity is shown for any BMS channel and $$\beta < 0.5$$ with $$\lambda \approx 0.631$$ . 

   more » « less
3. Deepcode: Feedback Codes via Deep Learning

   Kim, Hyeji; Jiang, Yihan; Kannan, Sreeram; Oh, Sewoong; Viswanath, Pramod (December 2018, Advances in neural information processing systems)

   The design of codes for communicating reliably over a statistically well defined channel is an important endeavor involving deep mathematical research and wideranging practical applications. In this work, we present the first family of codes obtained via deep learning, which significantly beats state-of-the-art codes designed over several decades of research. The communication channel under consideration is the Gaussian noise channel with feedback, whose study was initiated by Shannon; feedback is known theoretically to improve reliability of communication, but no practical codes that do so have ever been successfully constructed. We break this logjam by integrating information theoretic insights harmoniously with recurrent-neural-network based encoders and decoders to create novel codes that outperform known codes by 3 orders of magnitude in reliability. We also demonstrate several desirable properties in the codes: (a) generalization to larger block lengths; (b) composability with known codes; (c) adaptation to practical constraints. This result also presents broader ramifications to coding theory: even when the channel has a clear mathematical model, deep learning methodologies, when combined with channel-specific information-theoretic insights, can potentially beat state-of-the-art codes, constructed over decades of mathematical research. 

   more » « less
4. CRISP: Curriculum based Sequential neural decoders for Polar code family

   Hebbar, S.A.; Nadkarni, V.V.; Makkuva, A.V.; Bhat, S.; Oh, S.; Viswanath, P.. (July 2023, Proceedings of the 40th International Conference on Machine Learning)

   Polar codes are widely used state-of-the-art codes for reliable communication that have recently been included in the 5th generation wireless standards (5G). However, there still remains room for design of polar decoders that are both efficient and reliable in the short blocklength regime. Motivated by recent successes of data-driven channel decoders, we introduce a novel 𝐂ur𝐑𝐈culum based 𝐒equential neural decoder for 𝐏olar codes (CRISP). We design a principled curriculum, guided by information-theoretic insights, to train CRISP and show that it outperforms the successive-cancellation (SC) decoder and attains near-optimal reliability performance on the Polar(32,16) and Polar(64,22) codes. The choice of the proposed curriculum is critical in achieving the accuracy gains of CRISP, as we show by comparing against other curricula. More notably, CRISP can be readily extended to Polarization-Adjusted-Convolutional (PAC) codes, where existing SC decoders are significantly less reliable. To the best of our knowledge, CRISP constructs the first data-driven decoder for PAC codes and attains near-optimal performance on the PAC(32,16) code. 

   more » « less
5. Stopping Set Analysis for Polar-Polar Concatenated Codes Under BP Decoding

   https://doi.org/10.1109/TCOMM.2025.3581118  

   Zhu, Ziyuan; Siegel, Paul H (January 2025, IEEE Transactions on Communications)

   This paper investigates properties of polar-polar concatenated codes and their potential applications. We start by reviewing previous work on stopping set analysis for conventional polar codes, which we extend in this paper to concatenated architectures. Specifically, we present a stopping set analysis for the factor graph of concatenated polar codes, deriving an upper bound on the size of the minimum stopping set. To achieve this bound, we propose new bounds on the size of the minimum stopping set for conventional polar code factor graphs. The tightness of these proposed bounds is investigated empirically and analytically. We show that, in some special cases, the exact size of the minimum stopping set can be determined with a time complexity of O(N), where N is the codeword length. The stopping set analysis motivates a novel construction method for concatenated polar codes. This method is used to design outer polar codes for two previously proposed concatenated polar code architectures: augmented polar codes and local-global polar codes. Simulation results with BP decoding demonstrate the advantage of the proposed codes over previously proposed constructions based on density evolution (DE). 

   more » « less