Flow characteristics analysis and modeling of M2M service

Machine to Machine (M2M) communications are believed to be the most promising services in foreseeable future. However, characteristics of M2M service and its mathematical model have been rarely researched up till now, which is the foundation of exploring the service nature based on queuing theory (our undergoing research project). In this paper, we carefully analyze the characteristics of several kinds of typical M2M Small Data Service, and design the service logic and traffic generation function on network simulation platform. We model the collected traffic with multi-state Markov Modulated Poisson Process (MMPP-n) which is suitable to model the traffic with different nature. Referencing LAMBDA algorithm [10], we can determine the parameters of MMPP-n. We design a program which produces traffic samples according to any defined MMPP-n. So, by comparing two sets of (H, Cv), one is derived from traffic data collected on simulation platform, the other is calculated from data samples generated by MMPP-n, we can determine the optimal mathematical fitting. We conclude that MMPP-3 is most suitable to model the M2M aggregate traffic.