Skip to contentSkip to conversation

what is your 5GHz rank modifier?

DISCUSSION James Prestwood 2 participants 3 messages May 6, 2021

lore
Share & tools
lore
#0 lore raw

Hi,

Several others in the past have asked about IWD's 5/2.4GHz ranking and explained that on their particular network 2.4GHz is heavily preferred. I have done some minimal testing and also noticed this. IWD heavily weights signal strength, and in my case the 5GHz network is significantly lower than 2.4, e.g. -71db vs -43db. Although the signal is worse I have found that I actually get higher throughput on 5GHz in many cases. IWD's own data rate calculation (theoretical) is coming up higher on 5GHz since its 802.11AC/VHT vs 802.11N on 2.4GHz. The issue boils down to IWD's significant weight on signal strength.

Signal strength is usually the best way for rank calculation especially when you have two networks using the same technology. For this reason its probably a bad idea to reduce the signal strength weights.

As we have suggested others, it seems most appropriate to modify IWD's 5GHz ranking specifically to handle this scenario. By default the 5Ghz rank modifier is unused (set to 1.0). I would like to know if anyone else has been playing with [Rank].BandModifier5Ghz, and found values that work for them as we are considering changing the default to better handle the average user.

Or maybe IWD handles your network just fine the way it is? Either way it would be nice to get some feedback from the community.

Thanks,
James

#1 lore raw

Hi James,

Quoted 4 lines

As we have suggested others, it seems most appropriate to modify IWD's 5GHz ranking specifically to handle this scenario. By default the 5Ghz rank modifier is unused (set to 1.0). I would like to know if anyone else has been playing with [Rank].BandModifier5Ghz, and found values

Sort of. BandModifier5Ghz is a blunt tool, mostly there to blindly force 5G or not.

Quoted 2 lines

that work for them as we are considering changing the default to better handle the average user.

I don't think changing the default is a good idea since it doesn't take throughput into account. Have you looked into tweaking the RANK_MAX_SUPPORTED_RATE_FACTOR and RANK_MIN_SUPPORTED_RATE_FACTOR (or really the spread between them)?

Just looking at the base rssi, the formula weighs the network with -43db about 2x higher than at -71db. That is a pretty significant difference to overcome. As you say, rssi might need to be given less weight or taken out of the equation entirely and have everything be based on the expected throughput.

Still, -71db is below our roaming threshold, so that is also something to consider.

Regards,
-Denis

#2 lore raw
Quoted 18 lines · Denis Kenzior

Hi James,

Quoted 8 lines

As we have suggested others, it seems most appropriate to modify IWD's
5GHz ranking specifically to handle this scenario. By default the 5Ghz
rank modifier is unused (set to 1.0). I would like to know if anyone
else has been playing with [Rank].BandModifier5Ghz, and found values

Sort of.  BandModifier5Ghz is a blunt tool, mostly there to blindly force 5G or not.

Quoted 3 lines

that work for them as we are considering changing the default to better
handle the average user.

Quoted 6 lines

I don't think changing the default is a good idea since it doesn't take
throughput into account.  Have you looked into tweaking the RANK_MAX_SUPPORTED_RATE_FACTOR and RANK_MIN_SUPPORTED_RATE_FACTOR (or really the
spread between them)?

I did play around with those but I found that the calculated 2.4 rate was close enough that these values needed to change more drastically than what I was expecting. But I didn't look too deep into this either. I think you're right, the spread between them may need to increase to give the effect we want.

Quoted 11 lines

Just looking at the base rssi, the formula weighs the network with - 43db about
2x higher than at -71db.  That is a pretty significant difference to overcome.
As you say, rssi might need to be given less weight or taken out of the equation
entirely and have everything be based on the expected throughput.

Still, -71db is below our roaming threshold, so that is also something to consider.

Yeah, this occurred to me as well. The roam threshold should probably be tweaked for 5Ghz.

Quoted 3 lines

Regards,
-Denis