Understanding the Traffic Causality for Low-Latency Human-to-Machine Applications

This letter presents our experimental study on the unique characteristics of human-to-machine (H2M) traffic and a novel bandwidth allocation scheme for low-latency H2M applications. For the first time, a high correlation between real-time control and feedback traffic, defined as traffic causality, is reported. Then, exploiting this causality, an artificial intelligence-facilitated interactive bandwidth allocation (AIBA) scheme is proposed to reduce the latency for converged H2M application delivery over access networks. Compared to existing schemes, AIBA pre-allocates bandwidth for feedback based on the control traffic forwarded by the central office and achieves a gain of up to 60% in expedited feedback delivery.