Did you do any more research on these antennas you can share? What solution did you end up utilizing? I’m also facing a similar issue with poor signal strength and looking for a solution. I have the device publishing the strength and quality as part of it’s normal data publish every minute and find the values to be quite variable. I can’t test directly in the field so didn’t want to buy the Taoglas TLS.01.305111 just yet.
A note on the % values you received. The actual values for strength are -121db to -25db and quality -24.5db to 0db. The reference document online doesn’t show the math of how this is mapped to 0 to 100% exactly. I haven’t checked the coding in the github repo though. With decibles being a log scale, even 2% may be a big difference (depending on that mapping). Personally, I’ll likely switch over the code to see the db just so I can get the raw measurement for comparison. Something you may want to look at too.