Ido Tal

How to Construct Polar Codes

A method for efficiently constructing polar codes is presented and analyzed. Although polar codes are explicitly defined, straightforward construction is intractable since the resulting polar bit-channels have an output alphabet that grows exponentially with the code length. Thus, the core problem that needs to be solved is that of faithfully approximating a bit-channel with an intractably large alphabet by another channel having a manageable alphabet size. We devise two approximation methods which “sandwich” the original bit-channel between a degraded and an upgraded version thereof. Both approximations can be efficiently computed and turn out to be extremely close in practice. We also provide theoretical analysis of our construction algorithms, proving that for any fixed ε > 0 and all sufficiently large code lengths n, polar codes whose rate is within ε of channel capacity can be constructed in time and space that are both linear in n.

List decoding of polar codes

We describe a successive-cancellation list decoder for polar codes, which is a generalization of the classic successive-cancellation decoder of Arikan. In the proposed list decoder, up to L decoding paths are considered concurrently at each decoding stage. Simulation results show that the resulting performance is very close to that of a maximum-likelihood decoder, even for moderate values of L. Thus it appears that the proposed list decoder bridges the gap between successive-cancellation and maximum-likelihood decoding of polar codes. The specific list-decoding algorithm that achieves this performance doubles the number of decoding paths at each decoding step, and then uses a pruning procedure to discard all but the L “best” paths. In order to implement this algorithm, we introduce a natural pruning criterion that can be easily evaluated. Nevertheless, straightforward implementation still requires O(L · n2) time, which is in stark contrast with the O(n log n) complexity of the original successive-cancellation decoder. We utilize the structure of polar codes to overcome this problem. Specifically, we devise an efficient, numerically stable, implementation taking only O(L · n log n) time and O(L · n) space.

List Decoding of Polar Codes

We describe a successive-cancellation list decoder for polar codes, which is a generalization of the classic successive-cancellation decoder of Arıkan. In the proposed list decoder, L decoding paths are considered concurrently at each decoding stage, where L is an integer parameter. At the end of the decoding process, the most likely among the L paths is selected as the single codeword at the decoder output. Simulations show that the resulting performance is very close to that of maximum-likelihood decoding, even for moderate values of L. Alternatively, if a genie is allowed to pick the transmitted codeword from the list, the results are comparable with the performance of current state-of-the-art LDPC codes. We show that such a genie can be easily implemented using simple CRC precoding. The specific list-decoding algorithm that achieves this performance doubles the number of decoding paths for each information bit, and then uses a pruning procedure to discard all but the L most likely paths. However, straightforward implementation of this algorithm requires Ω(Ln2) time, which is in stark contrast with the O(n log n) complexity of the original successive-cancellation decoder. In this paper, we utilize the structure of polar codes along with certain algorithmic transformations in order to overcome this problem: we devise an efficient, numerically stable, implementation of the proposed list decoder that takes only O(Ln logn) time and O(Ln) space.

Hardware architectures for successive cancellation decoding of polar codes

The recently-discovered polar codes are widely seen as a major breakthrough in coding theory. These codes achieve the capacity of many important channels under successive cancellation decoding. Motivated by the rapid progress in the theory of polar codes, we propose a family of architectures for efficient hardware implementation of successive cancellation decoders. We show that such decoders can be implemented with O(n) processing elements and O(n) memory elements, while providing constant throughput. We also propose a technique for overlapping the decoding of several consecutive codewords, thereby achieving a significant speed-up factor. We furthermore show that successive cancellation decoding can be implemented in the logarithmic domain, thereby eliminating the multiplication and division operations and greatly reducing the complexity of each processing element.

On the construction of polar codes

We consider the problem of efficiently constructing polar codes over binary memoryless symmetric (BMS) channels. The complexity of designing polar codes via an exact evaluation of the polarized channels to find which ones are “good” appears to be exponential in the block length. In [3], Tal and Vardy show that if instead the evaluation if performed approximately, the construction has only linear complexity. In this paper, we follow this approach and present a framework where the algorithms of [3] and new related algorithms can be analyzed for complexity and accuracy. We provide numerical and analytical results on the efficiency of such algorithms, in particular we show that one can find all the “good” channels (except a vanishing fraction) with almost linear complexity in block-length (except a polylogarithmic factor).

Hardware Implementation of Successive-Cancellation Decoders for Polar Codes

The recently-discovered polar codes are seen as a major breakthrough in coding theory; they provably achieve the theoretical capacity of discrete memoryless channels using the low-complexity successive cancellation decoding algorithm. Motivated by recent developments in polar coding theory, we propose a family of efficient hardware implementations for successive cancellation (SC) polar decoders. We show that such decoders can be implemented with O(N) processing elements and O(N) memory elements. Furthermore, we show that SC decoding can be implemented in the logarithmic domain, thereby eliminating costly multiplication and division operations, and reducing the complexity of each processing element greatly. We also present a detailed architecture for an SC decoder and provide logic synthesis results confirming the linear complexity growth of the decoder as the code length increases.

Flexible and Low-Complexity Encoding and Decoding of Systematic Polar Codes

In this paper, we present hardware and software implementations of flexible polar systematic encoders and decoders. The proposed implementations operate on polar codes of any length less than a maximum and of any rate. We describe the low-complexity, highly parallel, and flexible systematic-encoding algorithm that we use and prove its correctness. Our hardware implementation results show that the overhead of adding code rate and length flexibility is little, and the impact on operation latency minor compared with code-specific versions. Finally, the flexible software encoder and decoder implementations are also shown to be able to maintain high throughput and low latency.

Constructing polar codes for non-binary alphabets and MACs

Consider a channel with an input alphabet that is finite but not necessarily binary. A method for approximating such a channel having a large output alphabet size by a degraded version of it having a smaller output alphabet size is presented and analyzed. The approximation method is used to construct polar codes for both single-user and multiple-access channels with prime input alphabet sizes.

Channel upgrading for semantically-secure encryption on wiretap channels

Bellare and Tessaro recently introduced a new coding scheme, based on cryptographic principles, that guarantees strong security for a wide range of symmetric wiretap channels. This scheme has numerous advantages over alternative constructions, including constructions based on polar codes. However, it achieves secrecy capacity only under a certain restrictive condition. Specifically, let V be the main channel (from Alice to Bob) and let W be wiretap channel (from Alice to Eve). Suppose that W has a finite output alphabet y, and let X and Y denote the input and output of W, respectively. Then the rate of the Bellare-Tessaro coding scheme is at most I(V) - Ψ(W), where I(V) is the capacity of V and Ψ(W) is given by Ψ(W) = log2|y|-H(Y|X). For symmetric channels, it is clear that Ψ(ΨW) ≥ I(W) with equality if and only if uniform input to W produces uniform output. Unfortunately, few symmetric DMCs satisfy this condition. In this paper, we show how the Bellare-Tessaro coding scheme can be extended to achieve secrecy capacity in the case where W is an arbitrary symmetric DMC. To this end, we solve the following problem. Given W and ε > 0, we construct another channel Q such that W is degraded with respect to Q while the difference between Ψ(<;3) and I(W) is at most ε. We also solve a closely related problem, where the output alphabet of Q is required to be of a given size M. In this case, we construct a channel Q that is equivalent to W, such that Ψ(<;3) is a small as possible. We furthermore extend these results, and thereby the applicability of the Bellare-Tessaro coding scheme, to channels with binary input and continuous output.

Bounds on the Rate of 2-D Bit-Stuffing Encoders

A method for bounding the rate of bit-stuffing encoders for 2-D constraints is presented. Instead of considering the original encoder, we consider a related one which is quasi-stationary. We use the quasi-stationary property in order to formulate linear requirements that must hold on the probabilities of the constrained arrays that are generated by the encoder. These requirements are used as part of a linear program. The minimum and maximum of the linear program bound the rate of the encoder from below and from above, respectively. A lower bound on the rate of an encoder is also a lower bound on the capacity of the corresponding constraint. For some constraints, our results lead to tighter lower bounds than what was previously known.