Coding for Segmented Edits with Local Weight Constraints

We study segmented edit channels where the channel input is divided into disjoint segments, and each segment suffers at most one error, either a deletion or an insertion. The model was first introduced by Liu and Mitzenmacher [2010] over the binary alphabet, and was extended for the q-ary alphabet by Abroshan et al. [2018]. In this work, we first propose an efficient construction for segments with less redundancy than previous works, hence significantly improving the redundancy over the entire sequence, for any q-ary alphabet where $q$ ≥ 3. Additionally, motivated by the applications of constrained codes in DNA-based data storage systems and energy harvesting communication channels, to reduce the probability of having errors, we also impose certain weight constraints in every segment instead of over the whole sequence. In particular, for DNA storage systems, besides error correction capability, our coding method guarantees that all segments in every codeword are almost GC-balanced.