Blocking Probability of Massive MIMO : What is the Capacity of a Massive MIMO IoT System?

We provide a methodology to evaluate the capacity of a Massive MIMO supported IoT system in which a large number of low cost low power IoT devices transmit and receive sporadic data. Numerous IoT devices are supported by a single cell Massive MIMO base station with maximum-ratio processing. Orthogonal reference signals (RSs) or pilots are assigned randomly to all the IoT devices for channel estimation purpose. The number of simultaneously active IoT devices follows Poisson distribution. Due to the tremendous number of IoT devices, orthogonal RSs are heavily reused, which severely degrades the receiver signal quality. One of the most important performance criteria for this kind of system is the blocking probability which shows the percentage of the outage IoT devices, and how we maintain the low blocking probability while supporting all the IoT devices simultaneously is particularly important question. We provide a theoretical analysis, and show that the RS reuse massive IoT devices can be divided into two groups based on the throughput performance, and also show that the blocking probability only depends on one of the groups. We show that increasing the number of service antennas and/or reducing the number of IoT devices can help to improve the performance of the blocking probability, however there is a regime in which the parameter adjustment helps little to improve the performance. Based on these factors, we provide a useful algorithm to improve the performance of blocking probability. A number of simulation results are also provided to validate the theoretical analysis.