An Emperical Traffic Model of M2M Mobile Streaming Services

Abstract

M2M video streaming service has been widely used. In order to evaluate the impact of the uplink traffic caused by M2M video streaming on the mobile network, it is necessary to establish the traffic model of the M2M mobile video services. There are two main types of traffic models in the existing video traffic model. The traffic model based on the self-similarity of video streaming is too complex to be simulate. The traffic model based on probability density did not consider the frame size of MPEG-4 videos. This paper selected several distributions with heavy-tail to fit several video sources. The video sources are selected according to the requirements of the M2M service aspects and the limitation of the mobile network resources. The distributions include Gamma distribution, Pareto distribution, lognormal distribution and Weibull distribution. After comparing the fitness figure, fitness figure deference ratio, mean frame size similarity, we find that lognormal distribution is able to reflect the characteristic of the video source frame size. Further, based on the analysis on the mean bit rate and frame rate, considering the recommendation of bandwidth for video from 3GPP, the frame rate of the traffic model is given in the paper. In summary, a traffic model including frame size and frame rate is presented in the paper. Fitting results show that the distribution presented in the paper is 0.07%~2% degree better than the distribution presented in 3GPP specification. The mean frame size of the model is 99% similar to the mean frame size of the video source. Further, the distribution used in traffic model is a commonly used mathematical distribution in major of simulation software.