Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Learning-Based Detection for Marker Codes over Insertion and Deletion Channels

Guochen Ma, Xiaopeng Jiao, Jianjun Mu, Hui Han, Yaming Yang

TL;DR

This work tackles marker-code detection over insertion/deletion channels when perfect channel state information (CSI) is unavailable or the channel model is unknown. It proposes two deep learning detectors: FBNet, a model-driven network obtained by unfolding the Forward-Backward algorithm and tying weights to form an RNN, and FBGRU, a data-driven approach using bi-GRU layers with CSI-agnostic inputs. Experiments show FBNet achieves CSI-robust performance close to the CSI-based FB algorithm, while FBGRU delivers superior robustness and performance when channel models are unknown and across weakly bursty scenarios. The results advance reliable marker-code detection for storage systems (e.g., DNA storage, racetrack memory) and open avenues for joint detection/decoding and transformer-based receivers in synchronization channels.

Abstract

Marker code is an effective coding scheme to protect data from insertions and deletions. It has potential applications in future storage systems, such as DNA storage and racetrack memory. When decoding marker codes, perfect channel state information (CSI), i.e., insertion and deletion probabilities, are required to detect insertion and deletion errors. Sometimes, the perfect CSI is not easy to obtain or the accurate channel model is unknown. Therefore, it is deserved to develop detecting algorithms for marker code without the knowledge of perfect CSI. In this paper, we propose two CSI-agnostic detecting algorithms for marker code based on deep learning. The first one is a model-driven deep learning method, which deep unfolds the original iterative detecting algorithm of marker code. In this method, CSI become weights in neural networks and these weights can be learned from training data. The second one is a data-driven method which is an end-to-end system based on the deep bidirectional gated recurrent unit network. Simulation results show that error performances of the proposed methods are significantly better than that of the original detection algorithm with CSI uncertainty. Furthermore, the proposed data-driven method exhibits better error performances than other methods for unknown channel models.

Deep Learning-Based Detection for Marker Codes over Insertion and Deletion Channels

TL;DR

This work tackles marker-code detection over insertion/deletion channels when perfect channel state information (CSI) is unavailable or the channel model is unknown. It proposes two deep learning detectors: FBNet, a model-driven network obtained by unfolding the Forward-Backward algorithm and tying weights to form an RNN, and FBGRU, a data-driven approach using bi-GRU layers with CSI-agnostic inputs. Experiments show FBNet achieves CSI-robust performance close to the CSI-based FB algorithm, while FBGRU delivers superior robustness and performance when channel models are unknown and across weakly bursty scenarios. The results advance reliable marker-code detection for storage systems (e.g., DNA storage, racetrack memory) and open avenues for joint detection/decoding and transformer-based receivers in synchronization channels.

Abstract

Marker code is an effective coding scheme to protect data from insertions and deletions. It has potential applications in future storage systems, such as DNA storage and racetrack memory. When decoding marker codes, perfect channel state information (CSI), i.e., insertion and deletion probabilities, are required to detect insertion and deletion errors. Sometimes, the perfect CSI is not easy to obtain or the accurate channel model is unknown. Therefore, it is deserved to develop detecting algorithms for marker code without the knowledge of perfect CSI. In this paper, we propose two CSI-agnostic detecting algorithms for marker code based on deep learning. The first one is a model-driven deep learning method, which deep unfolds the original iterative detecting algorithm of marker code. In this method, CSI become weights in neural networks and these weights can be learned from training data. The second one is a data-driven method which is an end-to-end system based on the deep bidirectional gated recurrent unit network. Simulation results show that error performances of the proposed methods are significantly better than that of the original detection algorithm with CSI uncertainty. Furthermore, the proposed data-driven method exhibits better error performances than other methods for unknown channel models.
Paper Structure (28 sections, 16 equations, 18 figures, 1 table)

This paper contains 28 sections, 16 equations, 18 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work on Deep Learning for Channel Detection and Decoding
  3. Black-box approaches
  4. Model-based approaches
  5. Our Contributions
  6. Preliminaries
  7. Channel Models Related to Insertions and Deletions
  8. IDS Channel
  9. ID-AWGN Channel
  10. Marker Codes and the Forward-Backward Algorithm
  11. FBNet: A Model-Driven Approach
  12. Deep Unfolding of the FB Algorithm
  13. The Design of FBNet
  14. The Construction of Inputs
  15. The Forward Cell
  16. ...and 13 more sections

Figures (18)

  • Figure 1: The IDS channel model.
  • Figure 2: The ID-AWGN channel model.
  • Figure 3: The system model of marker code: (a) Decoding with the FB algorithm; (b) Decoding with the proposed FBNet/FBGRU.
  • Figure 4: A deep FFNN architecture for calculating the forward quantities.
  • Figure 5: The structure of FBNet.
  • ...and 13 more figures