RS-RLNC: A Reinforcement Learning-Based Selective Random Linear Network Coding Framework for Tactile Internet

Abstract

Tactile Internet (TI) has very stringent networking requirements and the transport layer plays a crucial role in meeting these requirements. However, the transport layer has several inherent limitations (e.g., bufferbloat, incast issue, and head of line blocking) due to which the performance of the current transport layer solutions is not optimal. We advocate replacing the "store-and-forward" strategy in transport layer solutions with the "compute-and-forward" strategy. One way to implement the "compute-and-forward" strategy is random linear network coding (RLNC). This paper proposes a learning-based RLNC framework called RS-RLNC that utilizes network and receiver feedback to optimally select between block-RLNC and sliding-RLNC to improve overall network performance. We present a simulation-based performance evaluation of current transport layer solutions against the state-of-the-art RLNC and RS-RLNC in terms of throughput, latency, and decoding complexity. Delay is reduced by a factor of 8.5% and decoding complexity is improved up to 20% compared to the state-of-the-art. Simulation results indicate that RS-RLNC has the potential to meet the stringent requirements of TI applications. Additionally, we present three future directions outlining the evolution of RS-RLNC to enhance the transport layer for TI compatibility.