On the Impact of the Traffic Structure on Random Access Models for LTE-M and NB-IoT Systems

Internet-of-Things (IoT) applications are becoming a critical part for various industrial applications such enabling remote control and monitoring of complex industrial processes. Cellular massive machine type communications technologies (mMTC) including LTE-M and NB-IoT serve as enabler for these technologies. Differently from conventional Internet access technologies such as 5G New Radio (NR), the major performance degradation in LTE-M and NB-IoT stems from the random access phase, that is usually implemented by utilizing multi-channel ALOHA scheme. The conventional approach to analysis of this scheme assumes superposition theorem of point processes leading to constant rate approximation of the arrival process from user equipment (UE). However, this assumption often lead to inherent instability of the model that does not reflect realistic behavior. The aim of this paper is to characterize this behavior assessing the bounds for this approximation remain valid. We show that the time system remains in stable states and system throughput are sensitive to not only mean arrival rate but to higher order moments of the arrival statistics. Specifically, some distributions such as geometric ones leads to significant underestimation of the theoretical maximum system throughput.