Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dynamic Server Allocation Under Stochastic Switchover on Time-Varying Links

Hossein Mohammadalizadeh, Holger Karl

TL;DR

The paper tackles dynamic resource allocation to parallel queues under time-varying links and stochastic, state-dependent switching delays. It proposes ACI, a non-myopic, frame-based scheduler that amortizes switching costs and uses a Lyapunov drift framework to prove throughput-optimality within a scaled capacity region $\zeta\mathcal{C}$. The authors instantiate the model in a two-hop UAV-enabled FSO backhaul scenario, deriving practical forecasts for goodput and a switching-penalized dwell strategy, and show through simulations that ACI achieves high service utilization while enabling latency-focused variants (ACI-A, ACI-PA). The results highlight a fundamental throughput-latency trade-off governed by the urgency metric and frame length, with significant implications for mobile optical network scheduling. Overall, ACI provides a principled, tunable approach to robustly manage dynamic servers under stochastic switching in realistic wireless-backhaul settings.

Abstract

Dynamic resource allocation to parallel queues is a cornerstone of network scheduling, yet classical solutions often fail when accounting for the overhead of switching delays to queues with superior link conditions. In particular, system performance is further degraded when switching delays are stochastic and inhomogeneous. In this domain, the myopic, Max-Weight policy struggles, as it is agnostic to switching delays. This paper introduces ACI, a non-myopic, frame-based scheduling framework that directly amortizes these switching delays. We first use a Lyapunov drift analysis to prove that backlog-driven ACI is throughput-optimal with respect to a scaled capacity region; then validate ACI's effectiveness on multi-UAV networks with an FSO backhaul. Finally, we demonstrate how adapting its core urgency metric provides the flexibility to navigate the throughput-latency trade-off.

Dynamic Server Allocation Under Stochastic Switchover on Time-Varying Links

TL;DR

The paper tackles dynamic resource allocation to parallel queues under time-varying links and stochastic, state-dependent switching delays. It proposes ACI, a non-myopic, frame-based scheduler that amortizes switching costs and uses a Lyapunov drift framework to prove throughput-optimality within a scaled capacity region . The authors instantiate the model in a two-hop UAV-enabled FSO backhaul scenario, deriving practical forecasts for goodput and a switching-penalized dwell strategy, and show through simulations that ACI achieves high service utilization while enabling latency-focused variants (ACI-A, ACI-PA). The results highlight a fundamental throughput-latency trade-off governed by the urgency metric and frame length, with significant implications for mobile optical network scheduling. Overall, ACI provides a principled, tunable approach to robustly manage dynamic servers under stochastic switching in realistic wireless-backhaul settings.

Abstract

Dynamic resource allocation to parallel queues is a cornerstone of network scheduling, yet classical solutions often fail when accounting for the overhead of switching delays to queues with superior link conditions. In particular, system performance is further degraded when switching delays are stochastic and inhomogeneous. In this domain, the myopic, Max-Weight policy struggles, as it is agnostic to switching delays. This paper introduces ACI, a non-myopic, frame-based scheduling framework that directly amortizes these switching delays. We first use a Lyapunov drift analysis to prove that backlog-driven ACI is throughput-optimal with respect to a scaled capacity region; then validate ACI's effectiveness on multi-UAV networks with an FSO backhaul. Finally, we demonstrate how adapting its core urgency metric provides the flexibility to navigate the throughput-latency trade-off.
Paper Structure (16 sections, 53 equations, 7 figures, 2 tables)

This paper contains 16 sections, 53 equations, 7 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. System Model
  4. Queueing Formulation
  5. Network Architecture & Channel Characteristics
  6. Stochastic Switching Time Model
  7. Scheduling Approaches
  8. The Max-Weight Algorithm: A Theoretical Benchmark
  9. The ACI Scheduling Framework
  10. Throughput Optimality of the ACI Scheduler
  11. Constant-Factor Approximation Lemma
  12. Frame Drift Inequality and Negative Drift
  13. Guaranteed Switching and Starvation Avoidance
  14. ACI Urgency Metric Variations
  15. Evaluation
  16. ...and 1 more sections

Figures (7)

  • Figure 1: System model
  • Figure 2: Capacity region: with switching ($\zeta C$) vs without ($C$)
  • Figure 3: E2E channel PDF and outage of a single slave
  • Figure 4: Slave UAV mobility and switching dynamics
  • Figure 5: Overall delay CDF for ACI variants
  • ...and 2 more figures