From Stochastic Foundations to Emerging Frontiers: A Comprehensive Survey on Packet Jitter Modeling, Measurement, and Metrics in IP Networks

Abstract

Packet delay variation (jitter) affects the reliability and perceived quality of real-time IP services, but its models and metrics remain fragmented across layers and technologies. This structured survey synthesizes research published from 1990 through August 2026 in three methodological families: stochastic and analytical foundations, simulation and empirical approaches, and cross-layer application-aware methods. The evidence shows a progression from single-queue formulations to end-to-end, heavy-tailed, self-similar, and data-driven models. A reproducible illustration using 1440 RIPE Atlas measurement windows demonstrates why metric choice matters: the 99th percentile of absolute consecutive RTT differences was 8.17 ms, excess-over-minimum RTT reached 52.36 ms, and an RFC 3550-style recursive estimator peaked at only 6.14 ms. The main unresolved problems are unified modeling across heterogeneous wired–wireless paths, comparable high-resolution measurement, cross-layer QoE prediction, reproducibility, and timing-channel security. Priority directions are distribution-aware reporting, open benchmark traces, hybrid analytical–learning models, and stable closed-loop control for 5G/6G, TSN/DetNet, SDN/NFV, and edge networks.