Balancing Coding Overhead to Improve Sliding Window RLNC Performance

Abstract:
Sliding Window Random Linear Network Coding (SW RLNC) is an enabler of Ultra Reliable Low Latency Communications (URLLC) and could be utilized in several emerging networking scenarios, such as the Tactile Internet. While SW RLNC is a quite mature technique, there have been limited efforts to explore how key coding parameters affect the reliabilitydelay performance and the associated coding overhead. In this work, we focus on the impact of two massively important coding parameters of SW RLNC, namely, the code rate, which defines the level of injected redundancy in the original data stream, and the coding window size, which refers to the number of packets considered during encoding. We delve into the interplay of these parameters and showcase the importance of fine-tuning both parameters at the same time. Through extensive simulation scenarios, we highlight that it is possible to trade off code rate for the coding window size and vice versa without affecting the delivery-delay performance. More importantly, we also show that it is possible to trade coding redundancy for a limited increase in coding complexity. We, then, shift our attention on how SW RLNC can serve as a reliability-enhancing solution in several real-life scenarios in the domain of Tactile Internet through an extensive set of simulations. We showcase that the examined trade-off can enhance the bandwidth utilization without significant overhead in the delay and coding complexity performance.