Controlling Adaptive HARQ Erasure Coding for Real-Time Transport under Channel Model Mismatch (Accepted to ECRTS 2026)

Abstract:

Real-time cyber-physical systems require predictable reliability under hard deadline constraints. Standard protocols often optimize either for latency (e.g., UDP) or for full reliability (e.g., TCP, QUIC), with the latter potentially leading to unbounded retransmission delays. Existing protocols that offer partial reliability often use static FEC configurations that either waste bandwidth or fail to operate reliably under dynamic or challenging channel conditions. A fundamental challenge is the model mismatch problem: real-time adaptive erasure coding schemes require simple, efficiently interpretable models for channel estimation, yet these models systematically underfit the complex real-world channel dynamics. In this work, we present a contribution to the Predictably Reliable Real-time Transport protocol (PRRT) that compensates for underfitting channel estimation through a closed-loop control architecture, using packet debt as a metric for the extent to which the end-to-end erasure rate deviates from the application's target erasure rate. A compensated channel erasure rate is then fed into a novel constraint-aware, anytime incremental search algorithm that derives a near-optimal HARQ coding configuration satisfying the application's erasure and delay constraints. New to this search is its awareness of the encoding and decoding complexity of the resulting coding configurations. This allows devices to adaptively limit the search space to configurations within their computational capabilities, which is essential for constrained edge devices. We provide a new high-performance Rust reference implementation of PRRT and demonstrate that the system converges towards the target erasure rate, even under model mismatch, and quickly adapts to shifts in channel distribution.