Can Massive MIMO Support Uplink Intensive Applications?

TL;DR

This paper evaluates the capacity of cell-free Massive MIMO to support uplink intensive applications like connected vehicles, showing that ZF decoding significantly reduces the number of required access points and antennas.

Contribution

It derives new SINR expressions for cell-free Massive MIMO with ZF decoding and demonstrates its effectiveness in supporting uplink intensive applications with fewer resources.

Findings

ZF decoding enables support for uplink intensive applications with fewer access points.

Approximately 32-90 APs per km² suffice for urban and suburban scenarios.

Only 2 APs per km² are needed in rural areas using lower frequency bands.

Abstract

Current outdoor mobile network infrastructure cannot support uplink intensive mobile applications such as connected vehicles that collect and upload large amount of real time data. Our investigation reveals that with maximum-ratio (MR) decoding, it is theoretically impossible to support such applications with cell-free Massive MIMO, and it requires a very large number of service antennas in single cell configuration, making it practically infeasible; but with zero-forcing (ZF) decoding, such applications can be easily supported by cell-free Massive MIMO with very moderate number of access points (AP's), and it requires a lot more service antennas in single cell configuration. Via the newly derived SINR expressions for cell-free Massive MIMO with ZF decoding we show that uplink power control is unnecessary, and that with 10 MHz effective bandwidth for uplink data transmission, in urban and suburban morphologies, on the 2 GHz band, 90 single antenna AP's per square kilometer and 32 single antenna AP's per square kilometer are enough to support 18 autonomous vehicles respectively. In rural morphology, using 450 MHz band, only 2 single antenna AP's per square kilometer is enough.