Abstract
In this letter, we develop a framework for the analysis of massive multiple-input multiple-output (MIMO) networks with both spatial and temporal random arrival of traffic, which is important in the next generation network but has not been investigated before. Using stochastic geometry and queueing theory, we derive a tight closed-form approximation for the average number of packets successfully transmitted at any unit time slot and area, referred to as spatial mean packet throughput ( T ). To maximize T , we find that when the number of BS antennas ( M ) is much larger than the average number of devices per cell ( {\mathcal{ N}}_{\tt S} ), the BS should schedule as many devices as possible; however, when M is smaller or comparable with {\mathcal{ N}}_{\tt S} , scheduling too many devices will bring significant impairment on T , and the optimal number of scheduled devices should be obtained from our proposed solution. To further quantify this conclusion, we also present M/{\mathcal{ N}}_{\tt S} as the criteria to choose proper scheduling mechanisms. The effectiveness of our optimal solutions has been verified via simulations.