Mechanic’s guide to how a 5G Massive MIMO antenna works

Introduction

This is not intended to be an in-depth scientific analysis of 5G Massive MIMO, but an explanation of how an AAS antenna works, including beams and MIMO.

The antenna array

A Massive MIMO antenna is an array of 64 transceiver modules, each with a crossed dipole antenna array. The modules are usually wired up in pairs, with one module connected to each polarisation, meaning that each transmitter has two dipole antennas, although this has little relevance to the way the array works, it gives each transmitter some ‘gain’ and allows for phase steering of their output.

Each transmitter contributes to the Total Radiated Power (TRP) of the antenna. For instance, a 320 Watt TRP system runs 64x5W transmitters.

Every Resource Block is transmitted by every transmitter. The only difference is the way each resource block is phased to focus it in a particular direction. Yes, _every_ resource block is indivdually steered in _every_ transmitter. This means the whole array is transmitting every resource block at full array gain, even when the array would be sending data in multiple directions at the same time. This also means that whilst every RB is sent at full array gain, the EIRP in the direction the RB is going depends entirely on how many RB are going in that direction as a fraction of the TRP of the array. The array is maintaining a constant EIRP on a per-link basis, so long as the number of RBs committed to that connection remains the same.

An array is usually configured as a collection of traffic and broadcast ‘beams’ -preset directions of radiation. Generally, 7 or 8 traffic beams distributed across the width of a sector are pre-defined. This makes steering data a lot easier as ther site only needs to know which beam a UE is in - it doesn’t actually follow each user around. That’s more likely to happen in the 26GHz band, where there are fewer users and predefined beams are probably not suitable.

Note that every transmitter has a fixed power output per RB. Transmitting just a few RBs does NOT mean all the available power is going into those RBs. The power per RB is a function of the module power and the total number of RBs. This ensures path consistency as otherwise the EIRP would vary from frame to frame regardless of the path to whatever UEs were in service. UE power is variable and controlled to ensure that they all arrive at the receivers at about the same level, within what is possible under the circumstances. To improve reception of the array when a UE is edge of cell involves a combination of lowering the modulation scheme used to transmite a symbol, and/or narrowing the number of RBs being used, which would narrow the available bandwidth to increase the signal to noise ratio and allowing an increase in modulation level and therfore data throughput.

Yes, there is no randomisation of data across a carrier in 5G systems. That would be counterproductive, so when a user is assigned RBs for a transacion, they will be consecutive, limiting the channel width over which they are transmitted, and therefore the noise level.

All this means that it is very difficult to assess the actual EIRP of an array for EMI purposes. The current method of assuming TRP plus gain results in an EIRP figure unlikely to be attained in practice as it would mean that every RB would have to be transmitted on the same beam whilst the site is busy, which either implies that the site’s other beams are not busy at the same time, or that one particular user is consuming the entire site’s capacity. Either way, this is unlikely, and also implies that the site is in the wrong place.