Packet dispersion techniques and capacity estimationConstantinos

The packet pair technique estimates the capacity of a path (bottleneck bandwidth) from the dispersion (spacing) experienced by two back-to-back packets 5]]19]]13]. It has also been claimed that the dispersion of longer packet bursts (`packet trains') can measure the available bandwidth of a path 5]]9]]3]. This paper examines such packet pair and packet train dispersion techniques. We rst demonstrate that the dispersion of packet pairs in loaded paths follows a multimodal distribution , and discuss the queueing eeects that cause multiple local modes. The path capacity is not necessarily the global mode, and so it cannot be estimated using statistical procedures for the most common bandwidth range. The eeect of the probing packet size is also investigated, showing that the conventional wisdom of using maximum sized packet pairs is not optimal. On the contrary, if the probing packets of diierent packet pairs are of variable size, the sub-capacity local modes become wider and weaker. We then focus on the dispersion of long packet trains. Increasing the length of the packet train reduces the measurement variance, but the estimates converge to a value, referred to as Asymptotic Dispersion Rate (ADR), that is lower than the capacity, and that is not related with the available bandwidth. We show the eeect of cross traac on the dispersion of long packet trains, and derive analytic expressions for the ADR in certain path conngurations. Putting all pieces together, we present a capacity estimation methodology that has been implemented in a measurement tool called pathrate. We present capacity measurements of several Internet paths using pathrate, and evaluate its accuracy and its robustness to cross traac eeects.