hello,
I have a few questions about throughuput on 5GHz wireless links.
Does it depend on:
thanks
ksswd -
Throughput on 5Ghz is the same as any other band…
How good is the signal between the two physical points?
Is it noisy or, quiet RF space? The more noise the lower the quality likely of the received wireless data packets.
The longer the distance between two point the less RF energy the RX gets - meaning more than likely a lower TX/RX rate negoiated between the radios, i.e. 24mbps vice 54mbps.
In your example below " 5500 and 5520 " vs " 5500 and 5600 ", although technically the first frequency pair is sufficently seperated…I have found through experience that 2 channel (meaning 5540 instead 5520) yeild significantly better results… It is what the RX sensitivity and out of channel rejection capabilities of the wireless card you choose has for specifications.
AP Bridge <> station mode is faster than bridge or WDS. Nstreme on a link that is at least 15% loaded is faster than any of the previous. If your link in Nstreme falls below about 10% things can get ‘funny’ on packet delay because of the way Nstreme wants ‘full’ packets to send.
More links in a PtMP senario - lower throughput is not necessarily the case. It all depends on the capabilities of the AP and the CPEs attached wirelessly. If you have good signal strength between the units, minimual filtering/nat’ing etc in the AP (so it can concentrate on getting data to/from the CPEs) then you can get upwards of 70mbps wireless throughput. Adding more CPEs, up to a point does, not decrease this amount of throughput. After a certain point (number of CPEs) then yes the number of CPEs do impact the overall throughput because it takes ‘X’ time to process data to TX to each CPE and of course still have time to RX from the CPEs…
Not exactly a straight answer to your questions but I’ve given you the basic information you were looking for.
thanks for the answer.
I asked that question because I have significant throughput problems on few of my links.
What do you mean by noisy RF space? Is it detectable by MT oS? Are “Scan” and “Frequency use” features enough to determine RF “quietness”?
Yes, I have also noticed 2 channel yield better throughput even up to 5 times. But should I seperate frequency per device or per physical location?
I mean I have two AP at one physical location, one with 4 cards and the other with 5 cards, giving 9 antennas in total.
What If I have the same frequency on two radios, each of them on different device?
Or perhaps it depends on direction antennas are set for?
Anyway below are throughput test I made. Both devices are installed at the same place. I include RX tests only as they gave odd results.
All radios are Atheros AR514 working in WDS mode , OS is MT 2.9.46. Wireless configuration is identical on all links, mostly default values are set. Only SSIDs and frequencies are different.
Device A (RB 532)
AP_1 (5520) - 1 link only, ss -72, RX - 15 mbit
AP_2 (5580) - 1 link only, ss -66, Rx - 15 mbit
AP_3 (5640) - 1 link only, ss -82, RX - 3 mbit
AP_4 (5680) - 1 link only, ss -62, RX - 25 mbit
Device B (Intel 800Mhz)
AP_1 (5520) - 1st link, ss -71, RX - 9 mbit
2nd link, ss -73, RX - 6 mbit
AP_2 (5560) - 1st link, ss -77, Rx - 4 mbit
2nd link, ss -78, RX - 3 mbit
3rd link, ss -72, RX - 8 mbit
AP_3 (5600) - 1st link, ss -70, RX - 20 mbit
2nd link, ss -78, RX - 1 mbit !
AP_4 (5640) - 1 link only, ss -65, RX - 7 mbit
AP_5 (5700) - 1 link only, ss -60, RX - 5 mbit
Shouldn’t links with the same signal strength have comparable throughput and what else can affect performance?
ksswd -
What do you mean by noisy RF space? Is it detectable by MT oS? Are “Scan” and “Frequency use” features enough to determine RF “quietness”?
Well MT OS does a fair job, but there are things that esacpe it - Trango 5Ghz units for example…
Well it looks like you need to do some reading on RF energy/signal propragation…
You have two devices and with 9 radios…all in the same physical area…
If it were me - none of them would be on the same channel. You need at least 1 meter / 3 feet of seperation between antennas. I would also have the antennas alternate between vertical and horizontal polarization. Cards in the same box - prudent wise it is better to put them in seperate bands, i.e., 5.1ghz, 5.4ghz and 5.8ghz. Since that is not possible in your case then…the cards that are physically closer together inside each unit need to be as far seperated in frequency as possible. So you might want to alternate between the bands, card on the ‘top’ of you stack, 5.1ghz (5.1 includes all the way to 5.38ghz), next card 5.8ghz, then something in the high 5.4ghz band. Continue until you have all the cards programmed in your first device. Do the same in the second device, only try to seperate the channels from device one by at least two channels, and try not to put that on an antenna right next to something in the same band from the first device.
With what you have up there right now - you have made your own noise and your throughput is suffering because of it. And yes - as a general rule similiar signal strengths do produce similiar throughput speeds. The variables are number of clients / noise in the environment.
Thom