I've been optimizing my network and there are a number of ways of doing this. The method I'm seeking more information and understanding on is where one fixes the MCS values for the downlink on the AP and uplink on the SM. What I have read and tested:
Fixing MCS on the AP for downlink and SM for uplink should result in a more robust link when the signal is low.
MCS 11 provides 2 x spatial streams. In my test setup, I achieve approx. 20mbps uplink throughput at it's available signal strength.
MCS 7 provides 1 x spatial streams and also provides approx. 20mbps uplink throughput at the same availble signal strength.
So my questions are...
Does setting the AP to a lower MCS level for downlink degrade the total capacity of the AP?
Does 1 x spatial stream degrade the total capacity of the AP if some SM's communicate with 1 x spatial stream for the uplink?
Does setting the AP to a lower MCS level for downlink degrade the total capacity of the AP?
Does 1 x spatial stream degrade the total capacity of the AP if some SM's communicate with 1 x spatial stream for the uplink?
Thoughts and comments appreciated.
Thanks, Tim.
1. YES
2. YES
In a PtMP system, you typically do not want to lower or assigned a fixed MCS on the AP. For PtMP clients or for both ends of a PtP link, where the MCS is distributed between two modulation levels, if you're trying to achieve more constant throughput and/or lower jitter, try setting a fixed modulation to whatever MCS value has a higher % distribution. You can look this value up in the Monitor -> Performance area of the radio.
I just want to be clear on some things. My example is PTMP.
The fixed rate on the AP and SM is a max fixed rate so it will negotiate lower. I'm assuming that very rarely will customers in this test reliably go higher than MCS 14, so the AP has this fixed max rate – no doubt this will lower AP capacity on the downlink but thats not degredation. Is this correct?
Also on the SM, lowering the uplink MCS is a common method of improving uplink stability for the customer. The SM will never modulate higher, consistently, due to it's signal strength on the uplink, so setting the fixed max saves airtime if it were to attempt to negotiate higher from time to time. Weaker signal strength for SM's across the AP will reduce the overal performance of the AP. Is this correct?
I think the most important question is... Will a lower MCS consume the same amount of airtime as a higher modulation? To answer my own question (please correct me if I'm wrong), I believe it does, however this is split between downlink and uplink as part of the TDD scheduler. So for example...
10 customers at a lower MCS that yields 2meg each on uplink will have a total of 20meg on the uplink.
10 customers at a higher MCS that yields 5meg each on the uplink will have a total of 50meg on the uplink.
Both 1 and 2 above use the same amount of airtime on the uplink.
If the customers in 1 and 2 were combined, they would only affect the airtime on the uplink because of their differing MCS. That would reduce uplink capacity, but it would not affect downlink capacity.
Degredation of capacity will only show when the AP is under load.
The above does not take into consideration the affect of interference and other factors that would affect capacity.
My question back at you is why are you doing all this? We run many ePMP AP's and have many clients and we never set a fixed/max MCS rate on anything. For 99% of the situations the ePMP schedular does a very good job at handling clients with bad SnR/modulation. Are you experiencing specific issues that are making you think you need to used fixed/max MCS? The only times we ever play with using a fixed/max MCS are on PtP links, typically when we're dealing with noise and retransmissions that are causing latency and/or modulation flapping... and this is a last ditch effort after we've tried everything else (different channel and channel width, higher gain antennas, etc). We really really avoid setting a fixed/max MCS rate.
I'm not sure what the difference is between 'lower AP capacity' and 'degrade the total capacity of the AP'. To me these seem like that same things. By setting a lower fixed/max MCS rate on the AP, you will reduce overall AP capacity. Lower modulations increase frame utilization and will decrease capacity. It takes more airtime to send the same amount of data over MCS14 then it does MCS15.
Yes, having clients with poor SnR and poor modulation will reduce the overall AP capacity... BUT ePMP has a airtime fairness algorithm that helps reduce the impact that poor clients have on overall AP performance. You'll see this in effect by clients that have poor SnR/modulation having higher latency and jitter over their good peers during periods of high AP frame utilization.
In addition, when using flexible TDD framing you can further hide the effects of poor clients and also help increase their performance... BUT as mentioned, this is only a band-aid hiding the bad apples as the scheduler is robbing frame time from either the uplink or downlink dynamically as needed by the client, which in turn will reduce overall capacity.
By using fixed ratio TDD framing you will now have much more accurate bandwidth results with bad clients when doing speed tests and link evaluations as they're no longer able to rob a good client's frame time and airtime fairness further reduces their impact on overall AP capacity/quality.
Ideally you want to be maintaining high SnR and modulation across the AP and using fixed ratio TDD framing as you will 'paint yourself into a corner' with flexible and bad clients if you ever need to use fixed framing in the future (like using GPS sync to reduce guard bands between adjacent AP's and/or using back to back frequency reuse).
Hi Eric. Apologies for not responding earlier. Somehow missed this.
Thank you for your experiential input. The scheduler is super efficient and you make a very good point there. The airtime fairness is also another very relevant point to consider, and it's enlightening to be made aware of it's effectiveness. Pretty much answers my questions. To sum up for others:
You can use lower uplink modulations to sustain better connections:
However doing so will reduce the overall capacity of the radio.
This may not be noticeable as the airtime fairness on the AP will spread airtime evenly.
By keeping the modulations dynamic the radios should handle themselves.
Use only with clients that have bad uplink SNR and where you're getting disconnects in environements with odd interference symptoms.
In most instances never change the downlink modulation on the AP.
Changing modulation on PtP links is more feasible as opposed to doing so on PtMP links.
Many thanks for your input Eric. Please mark as resolved if you agree.