Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Adaptive Receiver-Side Scheduling for Smooth Interactive Delivery

Michael Luby

TL;DR

The paper tackles jitter in interactive streaming by introducing receiver-side scheduling that operates entirely on the receiver clock. It proposes Adaptive Delay Control (ADC), which maintains an adaptive offset $D$ to map sender timestamps $S_n$ to a stable release timeline, updating asymmetrically to track the upper envelope of delay using $X_n = A_n - (S_n + D)$. Enhancements such as Quantized ADC (QADC), a neutral band, and a bounded in-order guard improve cadence stability and sequencing, and the integration into the BitRipple Tunnel (BRT) demonstrates real-world viability. Evaluations on cloud-gaming workloads show substantial reductions in large jitter excursions and tighter inter-release intervals, suggesting broad applicability to other interactive pipelines and transport stacks.

Abstract

Interactive applications such as cloud gaming, XR streaming, and real-time inference depend on data objects arriving at a steady cadence. In practice, network delay variation and recovery dynamics at the receiver distort this cadence even when transports deliver all packets correctly, which produces visible jitter, stalls, and unstable playback. We present a lightweight receiver-side scheduling approach that regularizes release timing after recovery. The scheduler maintains an adaptive estimate of effective path delay and adjusts release times asymmetrically, responding quickly to late arrivals and only gradually to early ones. This upper-envelope behavior keeps release aligned with recent delay peaks and maintains smooth playback with minimal added latency. The scheduler runs entirely on the receiver clock and requires no feedback or synchronization. As a concrete example, we integrate receiver-side scheduling into the BitRipple Tunnel (BRT) overlay, an application-layer software system that forwards traffic without altering the underlying transport protocol. Within BRT, the scheduler functions as an independent module that regulates delivery timing for forwarded objects. Evaluating BRT with receiver-side scheduling on a cloud-gaming workload shows that the scheduler removes virtually all large jitter excursions and yields tightly clustered release intervals that improve visible smoothness. Broader latency improvements arise from the behavior of the full BRT overlay. Receiver-side scheduling can also be integrated modularly into transport stacks such as TCP, QUIC, WebRTC, UDP, or RTP, which are natural deployment points for future work.

Adaptive Receiver-Side Scheduling for Smooth Interactive Delivery

TL;DR

The paper tackles jitter in interactive streaming by introducing receiver-side scheduling that operates entirely on the receiver clock. It proposes Adaptive Delay Control (ADC), which maintains an adaptive offset to map sender timestamps to a stable release timeline, updating asymmetrically to track the upper envelope of delay using . Enhancements such as Quantized ADC (QADC), a neutral band, and a bounded in-order guard improve cadence stability and sequencing, and the integration into the BitRipple Tunnel (BRT) demonstrates real-world viability. Evaluations on cloud-gaming workloads show substantial reductions in large jitter excursions and tighter inter-release intervals, suggesting broad applicability to other interactive pipelines and transport stacks.

Abstract

Interactive applications such as cloud gaming, XR streaming, and real-time inference depend on data objects arriving at a steady cadence. In practice, network delay variation and recovery dynamics at the receiver distort this cadence even when transports deliver all packets correctly, which produces visible jitter, stalls, and unstable playback. We present a lightweight receiver-side scheduling approach that regularizes release timing after recovery. The scheduler maintains an adaptive estimate of effective path delay and adjusts release times asymmetrically, responding quickly to late arrivals and only gradually to early ones. This upper-envelope behavior keeps release aligned with recent delay peaks and maintains smooth playback with minimal added latency. The scheduler runs entirely on the receiver clock and requires no feedback or synchronization. As a concrete example, we integrate receiver-side scheduling into the BitRipple Tunnel (BRT) overlay, an application-layer software system that forwards traffic without altering the underlying transport protocol. Within BRT, the scheduler functions as an independent module that regulates delivery timing for forwarded objects. Evaluating BRT with receiver-side scheduling on a cloud-gaming workload shows that the scheduler removes virtually all large jitter excursions and yields tightly clustered release intervals that improve visible smoothness. Broader latency improvements arise from the behavior of the full BRT overlay. Receiver-side scheduling can also be integrated modularly into transport stacks such as TCP, QUIC, WebRTC, UDP, or RTP, which are natural deployment points for future work.

Paper Structure

This paper contains 54 sections, 78 equations, 6 figures, 1 table, 3 algorithms.

Table of Contents

  1. Introduction
  2. Contributions.
  3. Related Work
  4. Model and Problem Statement
  5. Design goals.
  6. Adaptive Receiver-Side Scheduling
  7. State, Variables, and Initialization
  8. Asymmetric Update
  9. Why asymmetry helps.
  10. Optional neutral band.
  11. Release Scheduling
  12. Illustrative Behavior
  13. System View
  14. System View
  15. Enhanced ADC
  16. ...and 39 more sections

Figures (6)

  • Figure 1: Recovery times versus smoothed release times.
  • Figure 2: Receiver-side scheduling sits after recovery and before release to the application.
  • Figure 3: Quantized scheduling with midpoint hysteresis.
  • Figure 4: Bounded in-order guard window.
  • Figure 5: Playback buffer deviation events for a cloud-gaming trace with and without smoothing.
  • ...and 1 more figures