Analysis and simulation of WF2Q+ based schedulers: comparisons, compliance with theoretical bounds and influence on end-to-end delay jitter

The statistical multiplexing of non-fixed-size packet flows over a network interface is implemented by different algorithms, which attempt to approximate the GPS "fluid model". One of the most effective proposals is WF2Q, which is optimal but complex, and resulted in different implementations: WF2Q+ and S-SPFQ. The former is not exactly specified by any algorithm and allows a number of implementations: this article aims to clarify the differences among them, with theoretical and simulative approaches. We implemented a discrete-event simulator which has been used to verify how these implementations match the WF2Q properties. The article ends with a discussion on the impact of the scheduler fairness property on delay-jitter.