The research team assessed the performance of a 128-element massive MIMO system operating at 3.5GHz and the trials indicated that this technology could offer spectrum efficiency figures in excess of 100 bits/s/Hz, improving upon the capacity of today’s long term evolution (LTE) systems a factor of ten.
The experimental set up used the National Instruments software-defined radio (SDR) platform that is being used to define 5G. The researchers programmed FPGAs using NI's LabVIEW communications system design and MIMO application framework.
The goals of the trials were to test massive MIMO spatial multiplexing indoors and improve the understanding of massive MIMO radio channels under mobile conditions with untethered devices. The trials were conducted in a large indoor hall mimicking a stadium environment and outdoors within the Adastral Park campus.
Initial experiments took place in BT's large exhibition hall and used 12 streams in a single 20MHz channel to show the real-time transmission and simultaneous reception of ten unique video streams, plus two other spatial channels demonstrating the full richness of spatial multiplexing supported by the system.
The system was also shown to support the simultaneous transmission of 24 user streams operating with 64QAM on the same radio channel with all modems synchronising over-the-air. It is believed that this is the first time such an experiment has been conducted with truly untethered devices, from which the team were able to infer a spectrum efficiency of just less than 100bit/s/Hz and a sum rate capacity of about 2Gbit/s in this single 20MHz wide channel.
In addition to the indoor trials, a series of outdoor experiments were conducted with the array at a height of around 20 meters. This enabled far-field array characterisation, multi-element handset performance as well as experiments to improve the understanding of the massive MIMO radio channel under mobile conditions to be carried out.
It is expected that massive MIMO, which offers full spatial multiplexing – where multiple data streams are