Research Overview - dcomp-leris/CGSynth GitHub Wiki

Research Overview

CGSynth: Cloud Gaming Synthesizer — SIGCOMM 2025 Demo.

Abstract

Cloud gaming's unique network traffic is challenging to reproduce. CGSynth generates realistic, configurable synthetic cloud gaming (CG) traffic. It captures real CG patterns and allows deterministic reproduction with user-defined flow/packet parameters. It employs a GRU for order-preserving timestamp generation and AI-based frame interpolation (RIFE) for realistic payloads. It integrates QoE evaluation (PSNR, SSIM, LPIPS, VMAF, MOS) to validate synthetic traffic's video quality and responsiveness against real streams.

System Highlights

  • GRU-based timestamp generator
  • RTP/PCAP payload pipeline
  • RIFE-based interpolation and degradation controls
  • Objective/subjective QoE validation
⚠️ **GitHub.com Fallback** ⚠️