Grant-Free Access for mMTC: A Performance Analysis Based on Number of Preambles, Repetitions, and Retransmissions

Abstract

Grant-free (GF) access is a multiple access mechanism being considered for supporting high user density in massive machine-type communication (mMTC). Since the GF schemes are predominantly contention based, a major challenge would be to handle user packet collisions. Additionally, users experiencing collision will retransmit their packets, leading to an increased load on an already dense network. Thus, the probability of successful transmission of an individual user and that of all users become the key performance indicators (KPIs) to analyze such systems. In this article, a GF uplink system is modeled for cases when the number of users is—1) fixed and 2) random (following a Poisson distribution). Analytical closed-form expressions for all-user and per-user success probabilities are derived in both the cases. Conventionally, GF access schemes have been analyzed using a fixed number of preambles. In this work, it is proposed to choose the number of preambles in the GF system such that the probability of success of at least “all but two” users is greater than or equal to 99%. It is shown that such a choice limits the maximum number of colliding users to be within four, thereby reducing the number of retransmissions. The analysis is extended to the case of transmission using multiple repetitions and numerical simulations are performed to validate the theoretical analysis. Finally, a tradeoff between the number of preambles, repetitions, and retransmissions is presented in terms of the per-user success probability, transmission latency, and energy consumption.