Document Type


Publication Date



Packet dispersion techniques, such as packet pair/train techniques have been commonly used to estimate bandwidth in wired networks. However, current packet dispersion techniques have been developed for wired network environments, which may lead to inaccurate results in wireless networks because of the variance in wireless capacities over short time scales. In this paper, we develop an analytical model to investigate the behaviors of packet dispersion in wireless networks. The packet dispersion model is validated using NS2, modified to support the 802.11 MAC layer rate adaptation. By utilizing of packet dispersion model, we clarify that the packet dispersion technique measures the effective capacity and achievable throughput of wireless networks instead of the capacity defined in wired networks. In addition, we analyze the performance of packets dispersion techniques in wireless networks, including the expected value and variance of estimation results and the interaction with channel conditions, such as packet sizes, link rate, bit error rate, and RTS/CTS access method. We show that the fluid traffic model is not applicable in over saturated wireless networks because of the probability based fairness across the nodes in wireless networks.