Will antenna improve my speeds?

[dalor]

I live in urban environment 200 metres from the cell tower my BS818 router connects to, but I have no windows facing the cell tower.
The best window is at 90 degrees angle to the cells. And there are lots of cells in band 1 and 3 on that tower which my router happily aggregates.

I get around 50 Mbps using a single-threaded speed test.

Can this be improved with a MIMO antenna?


Quiet period
   

Two download tests using testmy.net with the nearest server
   

[Seán]

The cell that you are connected to appears unusually quiet for the time of evening based on the RSRQ reading. It's definitely the quietest I've seen any cell at peak time.

Both the signal strength (RSRP) and quality (RSRQ) readings are excellent on this band. I suggest selecting just one band at a time to check the readings of both bands in case one is much weaker. If the RSRQ is around -6dB on the other band, then you'll probably not get any further speed improvement with an antenna.

The SINR however is unusually low for the RSRQ reading, which indicates cross polarisation interference. If you don't have any antenna attached, try moving or turning the router by 30 degrees to see if the SINR improves and repeat, e.g. start with the back facing out the window and repeat until the back faces the left or right. Don't worry about the RSRP figure as long as it doesn't drop below -100dBm on either band.

If there's no improvement in speed, there could be a bottleneck on the mast's backhaul such as with the microwave link if it's not connected with fibre backhaul.

[dalor]

Thanks, Sean. It really helped. I managed to squeeze another 10 Mbps by rotating the router.
Another question if I may - is there a way to avoid the DL speed "staircase" when, as I understand, additional bands are aggregated later in the download process.
See below for an example.
   

[Seán]

The aggregation bands are established or disconnected on demand at the cell end. While you can select bands are available, unfortunately, there's no way to force the cell to keep aggregated carriers established.

The stepped increase in bandwidth is possibly due to the TCP congestion control at the server end. The server will adjust the throughput to try matching the available bandwidth. When the download starts, your router is likely just connected on a single band at the time, so this is what the server will initially detect. As the download progresses and router connects to additional carriers, the server's TCP congestion protocol will affect how quickly it discovers and consumes the additional bandwidth.

One test you can try is download either the 100MB or 1GB file from the following test file page:
https://ams-nl-ping.vultr.com/

Vultr's test servers use the BBR TCP congestion protocol, which is very quick at adapting to fluctuating available bandwidth. Google's speed test also uses the BBR TCP congestion protocol.

[dalor]

Very useful, thanks. I've enabled BBR on my VPN server and that added 30-40% of bandwidth to that server, impressive!