[Paper Review] Joint Retransmission, Compression and Channel Coding for Data Fidelity under Energy Constraints

Chiara Pielli, Čedomir Stefanović+2 more|arXiv (Cornell University)|Jun 28, 2017

Energy Harvesting in Wireless Networks29 references3 citations

TL;DR

This paper proposes a joint retransmission, compression, and channel coding scheme for energy-harvesting sensors to maximize data fidelity under energy constraints. By combining lossy compression with hybrid ARQ and optimizing coding/compression rates via MDP and reinforcement learning, the scheme reduces average distortion—especially when retransmissions are used strategically, with performance gains dependent on the retransmission limit r and system parameters.

ABSTRACT

We consider a monitoring application where sensors periodically report data to a common receiver in a time division multiplex fashion. The sensors are constrained by the limited and unpredictable energy availability provided by Energy Harvesting (EH), and by the channel impairments. To maximize the quality of the reported data, the packets transmitted contain newly generated data blocks together with up to $r - 1$ previously unsuccessfully delivered ones, where $r$ is a design parameter; such blocks are compressed, concatenated and encoded with a channel code. The scheme applies lossy compression, such that the fidelity of the individual blocks is traded with the reliability provided by the channel code. We show that the proposed strategy outperforms the one in which retransmissions are not allowed. We also investigate the tradeoff between the value of $r$, the compression and coding rates, under the constraints of the energy availability, and, once $r$ has been decided, use a Markov Decision Process (MDP) to optimize the compression/coding rates. Finally, we implement a reinforcement learning algorithm, through which devices can learn the optimal transmission policy without knowing a priori the statistics of the EH process, and show that it indeed reaches the performance obtained via MDP.

Motivation & Objective

• To address the challenge of maintaining high data fidelity in energy-harvesting wireless sensor networks with intermittent energy supply.
• To jointly optimize compression and channel coding rates under stochastic energy availability and channel fading.
• To design a hybrid retransmission mechanism that improves reliability by retransmitting lost data blocks along with new ones.
• To develop an energy-efficient transmission policy that minimizes distortion while respecting energy constraints.
• To enable devices to learn optimal transmission strategies without prior knowledge of energy or channel statistics using reinforcement learning.

Proposed method

• Proposes a hybrid ARQ mechanism where each transmission frame includes a new data block and up to r−1 previously lost blocks, all compressed to fit a fixed slot size.
• Uses lossy compression to reduce data volume, enabling higher coding redundancy for error resilience.
• Models the system using a Markov Decision Process (MDP) to optimize compression and coding rates under energy and channel constraints.
• Employs a reinforcement learning algorithm (R-learning) to enable autonomous policy learning without prior knowledge of energy or channel statistics.
• Introduces a parametric energy dynamics model to capture the stochastic nature of energy harvesting.
• Analyzes distortion performance using realistic rate-distortion curves and outage probability for short-packet fading channels.

Experimental results

Research questions

• RQ1 How does allowing retransmissions of previously lost data blocks affect the average distortion in energy-constrained sensor networks?
• RQ2 What is the optimal trade-off between compression rate and channel coding rate under stochastic energy availability and channel fading?
• RQ3 How does the maximum number of retransmissions r impact the long-term distortion performance?
• RQ4 Can reinforcement learning effectively learn the optimal transmission policy without prior knowledge of energy or channel statistics?
• RQ5 Under what conditions is increasing r beneficial, and when does it lead to diminishing or even negative returns?

Key findings

• The proposed retransmission scheme consistently reduces average distortion compared to single-transmission schemes, especially when r ≥ 2.
• The distortion is minimized when r is chosen such that the optimal compression rate k⋆_r lies to the right of a critical threshold kV, which depends on r and system parameters.
• For small r, increasing r generally improves performance, but as r increases, the likelihood that k⋆_r falls below kV increases, making higher r suboptimal.
• When r > m (maximum compression level), k⋆_r = k⋆_r−1, and increasing r yields no benefit, indicating a saturation point.
• The reinforcement learning algorithm converges to the optimal policy derived via MDP, demonstrating its effectiveness in learning without prior statistical knowledge.
• The performance gain from retransmission is not proportional to r; instead, it depends critically on the discretization of compression levels (m) and energy availability.