Publications

Reliability is an inherent challenge for the emerging nonvolatile technology of racetrack memories, and there exists a fundamental relationship between codes designed for racetrack memories and codes with constrained periodicity. Previous works have sought to construct codes that avoid periodicity in windows, yet have either only provided existence proofs or required high redundancy. This paper provides the first constructions for avoiding periodicity that are both efficient (average-linear time) and with low redundancy (near the lower bound). The proposed algorithms are based on iteratively repairing windows which contain periodicity until all the windows are valid. Intuitively, such algorithms should not converge as there is no monotonic progression; yet, we prove convergence with average-linear time complexity by exploiting subtle properties of the encoder. Overall, we both provide constructions that avoid periodicity in all windows, and we also study the cardinality of such constraints.

We address the challenge of optimizing the capacity-achieving input distribution for a multinomial channel under the constraint of limited input support size, which is a crucial aspect in the design of DNA storage systems. We propose an algorithm that further elaborates the Multidimensional Dynamic Assignment Blahut-Arimoto (M-DAB) algorithm. Our proposed algorithm integrates variational autoencoder for determining the optimal locations of input distribution, into the alternating optimization of the input distribution locations and weights.

Recent advancements in DNA storage show that composite DNA letters can significantly enhance storage capacity. We model this process as a multinomial channel and propose an optimization algorithm to determine its capacity-achieving input distribution (CAID) for an arbitrary number of output reads. Our empirical results match a scaling law that determines that the support size grows exponentially with capacity. In addition, we introduce a limited-support optimization algorithm that optimizes the input distribution under a restricted support size, making it more feasible for real-world DNA storage systems. We also extend our model to account for noise and study its effect on capacity and input design.

Code release for the paper: “Input Optimization in the Composite DNA Storage Channel” by Adir Kobovich and Nir Weinberger.

Recent experiments have shown that the capacity of DNA storage systems may be significantly increased by synthesizing composite DNA letters. In this work, we model a DNA storage channel with composite inputs as a multinomial channel, and propose an optimization algorithm for its capacity achieving input distribution, for an arbitrary number of output reads. The algorithm is termed multidimensional dynamic assignment Blahut-Arimoto (M-DAB), and is a generalized version of the DAB algorithm, proposed by Wesel et al. [1] developed for the binomial channel. We also empirically observe a scaling law behavior of the capacity as a function of the support size of the capacity-achieving input distribution.

This paper presents a novel approach to address the constrained coding challenge of generating almost-balanced sequences. While strictly balanced sequences have been well studied in the past, the problem of designing efficient algorithms with small redundancy, preferably constant or even a single bit, for almost balanced sequences has remained unsolved. A sequence is ε(n)-almost balanced if its Hamming weight is between 0.5n±ε(n). It is known that for any algorithm with a constant number of bits, ε(n) has to be in the order of Θ(√n), with O(n) average time complexity. However, prior solutions with a single redundancy bit required ε(n) to be a linear shift from n/2. Employing an iterative method and arithmetic coding, our emphasis lies in constructing almost balanced codes with a single redundancy bit. Notably, our method surpasses previous approaches by achieving the optimal balanced order of Θ(√n). Additionally, we extend our method to the non-binary case considering q-ary almost polarity-balanced sequences for even q, and almost symbol-balanced for q=4. Our work marks the first asymptotically optimal solutions for almost-balanced sequences, for both, binary and non-binary alphabet.

Constrained coding is a fundamental field in coding theory that tackles efficient communication through constrained channels. While fixed constraints (e.g., a fixed set of substrings may not appear in transmitted messages) have a general optimal solution, there is increasing demand for supporting parametric constraints that are dependent on the message length and portray some property that the substrings must satisfy (e.g., no log (n) consecutive zeros). Several works have tackled such parametric constraints through iterative algorithms following the sequence-replacement approach, yet this approach requires complex constraint-specific properties to guarantee convergence through monotonic progression. In this paper, we propose a universal framework for tackling any parametric constraint problem with far fewer requirements, through a simple iterative algorithm. By reducing an execution of this iterative algorithm to an acyclic graph traversal, we prove a surprising result that guarantees convergence with efficient average time complexity even without requiring any monotonic progression. We demonstrate how to apply this algorithm to the run-length-limited, minimal Hamming weight, local almost-balanced Hamming weight constraints, as well as repeat-free and secondary-structure constraints. Overall, this framework enables state-of-the-art results with minimal effort.