Prediction of round trip delay for wireless networks by a two-state model

A method has been developed to predict the probability density function of the round-trip time (RTT) of a mobile/wireless network in an online manner by using data obtained from probe packets. In the present paper, we show the measurement results for the RTT by using LTE and Wi-Fi. On the basis of the results, we have built a two-state model. The states of the model correspond to a connected state, where the RTT is given by the shifted gamma distribution, and a disconnected state, where the packet is queued. The proposed method predicts the future probability density of the RTT by estimating the model parameters. A comparison with conventional methods revealed the proposed method can predict the probability density function more precisely with better stability.