PNC Enabled IIoT: A General Framework for Channel-Coded Asymmetric Physical-Layer Network Coding

Zhaorui Wang, Ling Liu, Shengli Zhang, Pengpeng Dong, Qing Yang, Taotao Wang

TL;DR

This paper introduces a novel framework for Physical-Layer Network Coding (PNC) in Industrial Internet of Things (IIoT) scenarios, specifically addressing the challenge of asymmetric information transmission between a controller and a robot. By employing a lattice-based encoding and decoding scheme, the authors enable the controller to transmit more data than the robot, reducing communication latency without sacrificing reliability.

What Problem Does This Paper Address?

In many Industrial IoT (IIoT) applications, a controller needs to communicate with a robot that is out of direct range, requiring a relay. A common scenario involves the controller sending more complex instructions or data than the robot, which might only send acknowledgments. Furthermore, the communication channel between the controller and the relay is often stronger than the one between the robot and the relay. Traditional PNC techniques, however, assume both users transmit the same amount of information to ensure successful decoding at the relay. This limitation prevents efficient utilization of the stronger controller-to-relay channel and leads to increased communication latency. The paper tackles the challenge of designing a PNC system that can handle this asymmetry in information quantity and channel strength, specifically focusing on how the relay can decode network-coded messages when the superimposed signals carry different amounts of information from each user.

Methods

The core innovation of this paper is a lattice-based encoding and decoding scheme for asymmetric PNC. Here's a breakdown:

  • Asymmetric Transmission: The controller (user B) and robot (user A) transmit simultaneously to the relay. The controller, leveraging its stronger channel and larger information payload, uses a higher modulation order (e.g., 8-QAM) compared to the robot (e.g., QPSK).
  • Lattice-Based Encoding: Both users encode their information into lattices. Crucially, they use different "lattice construction levels" (L_A and L_B). This allows them to effectively modulate different amounts of information. The higher the lattice level, the more information can be encoded. The controller uses a higher level (L_B > L_A) to transmit more data.
  • Level-by-Level Decoding at Relay: The relay decodes the superimposed signals from the controller and robot in a "level-by-level" manner. It first decodes the information from the lowest lattice level and then progressively decodes higher levels.
  • Power Shaping and Correction Signal: A challenge arises because the power shaping used to constrain the transmitted signal's power might not be compatible across different lattice levels when L_A ≠ L_B. To address this, the paper proposes using a "correction signal" transmitted by the controller. This signal helps align the power-shaped signals, ensuring successful decoding at the relay.
  • Polar Source Coding for Correction Signal: To minimize the latency introduced by transmitting the correction signal, the paper employs polar source coding to compress it. This significantly reduces the transmission time of the correction signal, especially when the channel coding rate at the controller's lattice level is high.
  • Dynamic Transmission Scheme: To further optimize performance, a dynamic scheme is proposed where the relay chooses between the asymmetric (ALEM) and a symmetric transmission scheme based on which one offers a shorter overall transmission time.

Datasets

Not discussed in the paper. The paper focuses on theoretical framework development and simulation-based evaluation.

Key Insights and Ideas

  • Lattice Codes as a Unifying Framework: The paper demonstrates the power of lattice-based coding for handling complex communication scenarios like asymmetric PNC. Lattices provide a flexible structure that can accommodate different modulation orders and information rates.
  • Level-by-Level Decoding is Key: The ability to decode superimposed signals in stages, based on lattice levels, is crucial for making asymmetric PNC work. This avoids the need for a single, complex decoder that must handle all information simultaneously.
  • Correction Signal Bridges the Gap: The introduction of a correction signal is an ingenious way to overcome the incompatibility issues arising from different power shaping and lattice levels. It ensures that even with asymmetric information, the decoded signals at the relay are meaningful.
  • Polar Coding for Efficiency: Using polar source coding to compress the correction signal is a smart optimization. It highlights how advanced coding techniques can be integrated to improve practical system performance by reducing overhead.
  • Trade-off Between Asymmetric and Symmetric: The paper acknowledges that the asymmetric scheme isn't always superior. The overhead of the correction signal can sometimes make it slower than a symmetric approach, leading to the proposal of a dynamic switching mechanism.

Key Results

  • Reduced Latency: The proposed asymmetric PNC scheme significantly reduces communication latency compared to traditional symmetric schemes, especially when the controller has substantially more information to transmit.
  • Improved Throughput: Simulations show that the asymmetric transmission scheme (ALEM) achieves higher throughput than symmetric schemes (SLEM/STEM) when the controller transmits more data, demonstrating the benefit of exploiting channel asymmetry. For instance, with KB = 639 bits, ALEM achieved a 15% throughput improvement over SLEM.
  • Effectiveness of Lattice-Based Approach: The lattice-based encoding and decoding scheme successfully enables the relay to decode network-coded messages even when the superimposed symbols contain different amounts of source information.
  • Correction Signal Mitigation: The correction signal mechanism effectively addresses the decoding failures caused by incompatible power shaping in asymmetric lattice constructions.
  • Compression Benefits: Polar source coding effectively reduces the transmission time of the correction signal, particularly when the channel coding rate at the controller's lattice level is high.

Limitations and Future Work

  • Correction Signal Overhead: While polar coding helps, the correction signal still introduces some overhead, which can, in certain scenarios (e.g., low coding rate at the controller's lattice level), lead to the asymmetric scheme being slower than a symmetric one. The paper addresses this with a dynamic transmission scheme.
  • Imperfect Channel Precoding: The paper investigates the performance penalty of imperfect channel precoding, showing that while some deviation is tolerable, significant offsets can impact throughput.
  • Decoding Complexity: The paper briefly discusses the decoding complexity, noting that it scales with the number of lattice levels and the complexity of the underlying channel codes. Further exploration of decoding properties and complexity for various channel codes is suggested.
  • Future Work: The authors suggest further exploration of the decoding properties of the lattice decoder, especially when the correction signal is not perfectly decoded. They also leave the analysis of decoding FER and coding gain in complicated received signal structures as a future research direction.

Summary updated

Abstract

This paper investigates the application of physical-layer network coding (PNC) to Industrial Internet of Things (IIoT) in which a controller and a robot are out of each other’s transmission range, and they exchange messages with the assistance of a relay. We particularly focus on a scenario where 1) the controller has more information to transmit than the robot; 2) the channel of the controller is stronger than that of the robot, and both users have nearly the same transmit power. To reduce the communication latency, we put forth an asymmetric PNC transmission scheme in which the controller transmits more information than the robot by exploiting its stronger channel gain in the uplink of PNC. However, the current channel-coded PNC requires the two users to transmit the same amount of source information in order to preserve the linearity of the two users’ channel codes at the relay for successful decoding. Therefore, a challenge in the asymmetric PNC transmission scheme is how to construct a channel decoder at the relay, considering that a superimposed symbol at the relay contains different amounts of source information from the controller and robot. To fill this gap, we propose a lattice-based encoding and decoding scheme in which the robot and controller encode and modulate their information in lattices with different lattice construction levels. The network-coded messages are decoded level-by-level in the lattice. Our design is versatile on that the controller and the robot can freely choose their modulation orders based on their channel power, and the design is applicable for arbitrary channel codes, not just for one particular channel code. The simulation results demonstrate the effectiveness of the proposed channel-coded asymmetric PNC transmission scheme.
BibTeX
@article{wang2022pnc,
  author = {Zhaorui Wang and Ling Liu and Shengli Zhang and Pengpeng Dong and Qing Yang and Taotao Wang},
  title = {{PNC Enabled IIoT: A General Framework for Channel-Coded Asymmetric Physical-Layer Network Coding}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2022},
  url = {https://openreview.net/forum?id=bcuW9ELAWn},
}
Figure