Mathematical and experimental analysis of limitations in creating artificial jitter for networked usability trials

Controlled usability trials are frequently desirable when assessing the impact of network latency, loss and jitter on highly interactive networked applications such as online games. This leads to a requirement for accurate (or at least predictable) emulation of IP level latency, loss and jitter on a localized network testbed. This paper reflects on the mathematical and experimental insights we gained when developing a controlled jitter network environment. We used FreeBSD’s kernel-resident dummynet module to introduce controlled jitter, but our results generalize to other tools that can introduce dynamically variable delays in an IP packet path. We expect these insights will stimulate further userexperience trials built around low-cost, unix-based networking tools.