Throughput-Optimal H-QMW Scheduling for Hybrid Wireless Networks With Persistent and Dynamic Flows

The well-known Queue-length-based MaxWeight scheduling algorithm (QMW) has been proved to be throughput-optimal for persistent flows only, which are long-lived with infinite traffic arrival. If the flows are dynamic ones, i.e., short-lived with finite data to transmit, QMW cannot guarantee queue stability. Given future wireless networks may support both persistent machine-to-machine flows and dynamic human-to-human flows, a Flow (File) Delay based MaxWeight scheduling algorithm (F-D-MW) has been shown to be throughput-optimal. However, new flows have to suffer a long start-up latency after arriving in the system. In this work, we present the definition of the capacity region for hybrid systems with the coexistence of persistent and dynamic flows. First, when a new arrival dynamic flow classification is known, we propose an online Hybrid Queue-length-based MaxWeight (H-QMW) scheduling algorithm, and then propose a more realistic adaptive H-QMW (A-H-QMW) scheduling algorithm for the system without the knowledge of the classification of flows. We prove that H-QMW can achieve throughput-optimality for hybrid systems. Performance evaluation not only validates the throughput-optimality of H-QMW and A-H-QMW in various types of networks but also reveals that H-QMW and A-H-QMW can achieve lower start-up and total latency for dynamic flows than F-D-MW.