Massive MIMO will almost certainly play a massive part in the 5G future. MIMO already plays a role in 4G, but the demands of a 5G network – one that is built to deliver high capacity and high speed data to a large number of subscribers all at once – have made the idea of arrays of tens or hundreds of antennas, splitting the signal workload between them, highly attractive to operators.
But Massive MIMO may not be appropriate in all cases. In fact some operators believe the use of mmWave spectrum will be important to making Massive MIMO deployable, as it will help keep antenna size and cost under control. Not surprisingly therefore a number of projects are now investigating the pros and cons of mmWave’s potential role – and alternatives to mmWave – in enabling affordable Massive MIMO.
For example a 5G R&D project being run by the universities of Bristol, UK and Lund, Sweden is focused on how to deploy massive antenna arrays without having to venture into the technically challenging mmWave frequencies – at least for now. The scientists involved have claimed a new world record in spectral efficiency, which they improved by 12 times using a 128-antenna system in 3.5 GHz spectrum. The design has already been tested, using these conventional microwave frequencies, using National Instruments’ LabView prototyping platform.
While spectrum above 20 GHz can support massive capacity within small cells, there are still many aspects of mmWave workings which are not fully understood in real deployment scenarios. In the meantime, the challenge will be to get hefty improvements out of sub-6 GHz bands, and perhaps to continue to use these as the mainstay of 5G networks well into the 2020s.
That will require greater efficiency because the microwave bands are becoming crowded, unlike the lightly occupied mmWave bands. That is the main focus of the Bristol/Lund project. “We see Massive MIMO as the most promising 5G technology,” said Ove Edfors [[i]], professor of radio systems at Lund University.
With this Massive MIMO design, up to 12 single-antenna clients can be supported, and each of those 12 shares a common 20 MHz radio channel. That approach delivered the “unprecedented” bandwidth efficiency of 79.4bps/Hz – the equivalent of 1.59Gbps in that one 20 MHz channel – during the tests.
The demo was conducted in the atrium of Bristol’s Merchant Venturers Building, with hardware provided as part of the Bristol Is Open programmable smart city platform. Lund has a similar platform so that the two institutions can carry out work in parallel.
“This activity reinforces our well established propagation and system modelling work by offering a new capability in model validation for Massive MIMO architectures,” Andrew Nix, head of Bristol’s CSN Group and the Dean of Engineering, said.
In another Massive MIMO/5G demo in midband spectrum, EE, working with Huawei, recently showed an end-to-end connection with a stable speed of 2.8Gbps, and a latency below 5ms. This test used a virtual core network and a 64×64 Massive MIMO unit, running in 3.5 GHz spectrum.
Several mobile network operators around the world have already begun rolling Massive MIMO technology out. How long they can do so affordably and with moderately sized antennas without resorting to mmWave spectrum will be interesting. After all, mmWave may be technically challenging but, put very simply, there is a lot of it available.