We hear a lot about 5G these days as competition between network equipment vendors, operators and chipsets manufacturers is heating up to be the one to impose its views on how 5G should be standardized. But how fast can 5G be?

speedtest-5g.jpg

Here is a summary of 2016 press releases found over the Internet which describe their 5G tests scnenarios and the maximum peak throughput that could be achieved (the summary is focusing on the spectrum used):

Date Country Operator Vendor Peak Speed Band, Bandwidth Technical Details
12-Feb Japan University Panasonic 100 Gbps 300 GHz Terahertz radios
18-Feb Sweden DoCoMo Ericsson 25 Gbps MU-MIMO, Beam Tracking
3-Jun USA Sprint Nokia 2.3 Gbps 73 GHz Beam Switching
6-Jun USA AT&T Ericsson 10 Gbps 15 GHz
15-Jun USA Sprint Ericsson 4 Gbps 15 GHz 50 feet, gNB to UE
11-Aug USA AT&T Ericsson 14 Gbps 15 GHz Multi-User MIMO
20-Sep USA T-Mobile Ericsson 12 Gbps 28 GHz 2-directional Beam Steering
20-Sep Australia Telstra Ericsson 20 Gbps ?, 800 MHz CA, 4×4 MIMO, 256 QAM
12-Oct Australia Vodafone Nokia 4.84 Gbps 4.5 GHz, 200 MHz 8×8 MIMO and 64 QAM
14-Oct Sweden Telia Ericsson 15 Gbps 15 GHz, 800 MHz
26-Oct Japan DoCoMo Fujitsu 11 Gbps 4.6 GHz Multiple Base Stations, 4 UEs
2-Nov Singapore StarHub Nokia 4.3 Gbps 3 to 30 GHz Nokia AirScale, 5G cmWave
12-Nov Korea LG Uplus Huawei 31 Gbps 28 GHz MIMO
16-Nov Australia Optus Huawei 35 Gbps 73 GHz Polar code
17-Nov Japan DoCoMo Huawei 11.29 Gbps 4.5 GHz, 200 MHz MIMO, Polar Code,
SCMA, 23 UEs, 64 TRX
19-Nov Japan DoCoMo Ericsson 10 Gbps 15 GHz MIMO

Although this summary focuses on throughput, in most of the cases, a latency of 1ms to ~5ms is also announced, which is a great improvement compared to 4G performance.

cmWave-mmWave.png

Disclaimer: these are theorical tests aimed at showing the maximum speed in a press release, the peak throughput often refers to the gNB total throughput with mutliple users, and in other cases, the user and the site are only of a few meters away.