Affordable Self-Tuning Smart Queue Management for Home Networks

Poorly managed buffers in network systems lead to significant delays in packet transmission, resulting in high latency and jitter that severely impact real-time applications and overall network efficiency. Traditional solutions, such as smart queue management (SQM) algorithms and buffer size reduction, are often unavailable in affordable off-the-shelf devices, and custom router setups require extensive technical knowledge. This paper introduces a cost-effective method to create a home network that utilizes SQM and performs self-tuning through a reinforcement learning (RL) approach, leveraging third-party open resources and affordable devices. By configuring an OpenWRT access point and applying RL settings, our solution effectively minimizes latency caused by bufferbloat. Evaluation results indicate that our RL-based approach can adapt to varying network conditions, offering a practical and cost-effective solution for optimizing network performance and mitigating bufferbloat in real-world scenarios.