Community discussions

 
syadnom
Member
Member
Topic Author
Posts: 405
Joined: Thu Jan 27, 2011 7:29 am

gps sync

Mon Sep 15, 2014 4:47 am

Lets get some facts straight.

The only Ubiquiti products with functional GPS sync are AirFiber5 and AirFiber24. Their Rocket M5-GPS *DOES NOT WORK*

The only products even close to 'tik and ubnt prices w/ functional GPS sync are Cambium ePMP.

GPS absolutely doesn't not allow channel re-use at full radio capacity. In the worst case, every same-channel synced AP will share the throughput of just 1 AP. This is true of AirFiber, Cambium, EVERYTHING, it's physics period. In *good* GPS sync, the radios will measure the interference of other radios and truly re-use timeslots when there is appropriate isolation (cambium) but this is *HARD* and is the kind of stuff that comes out of Motorola engineers when they are in charge of a project (Motorola's engineers are world-class, Motorola management is worthless).

GPS sync isn't completely necessary. Good antennas and good shield kits (RF-Armor!) can do miracles. It's a great feature in dense areas with a lot of devices for sure (again see cambium ePMP), but when you get to lighter density and rural areas, ubnt and 'tik perform as well if not better than ePMP.
 
ste
Forum Guru
Forum Guru
Posts: 1815
Joined: Sun Feb 13, 2005 11:21 pm

Re: gps sync

Mon Sep 15, 2014 8:16 am

Lets get some facts straight.

The only Ubiquiti products with functional GPS sync are AirFiber5 and AirFiber24. Their Rocket M5-GPS *DOES NOT WORK*

The only products even close to 'tik and ubnt prices w/ functional GPS sync are Cambium ePMP.
Yes. ePMP is still in an developing stage but respects regulations. So ATPC and DFS is implemented. False Radar detection is still a problem.
GPS absolutely doesn't not allow channel re-use at full radio capacity. In the worst case, every same-channel synced AP will share the throughput of just 1 AP. This is true of AirFiber, Cambium, EVERYTHING, it's physics period.
No. You can use same channels back2back. The APs do not interfere due to GPS and the CPEs reach the second AP from the back side.
The AP Antennas need a good Front2back ratio.
In *good* GPS sync, the radios will measure the interference of other radios and truly re-use timeslots when there is appropriate isolation (cambium) but this is *HARD* and is the kind of stuff that comes out of Motorola engineers when they are in charge of a project (Motorola's engineers are world-class, Motorola management is worthless).

GPS sync isn't completely necessary. Good antennas and good shield kits (RF-Armor!) can do miracles. It's a great feature in dense areas with a lot of devices for sure (again see cambium ePMP), but when you get to lighter density and rural areas, ubnt and 'tik perform as well if not better than ePMP.
The combination of all will bring the best results. Good shielding, good engineered radios and GPS.
Problem is the time. When current ePMP is fully functional it is old. It is based on 11n while we install
the first .ac connections with MT.
 
syadnom
Member
Member
Topic Author
Posts: 405
Joined: Thu Jan 27, 2011 7:29 am

Re: gps sync

Mon Sep 15, 2014 8:23 am

No. You can use same channels back2back. The APs do not interfere due to GPS and the CPEs reach the second AP from the back side.
The AP Antennas need a good Front2back ratio.
I said 'at full capacity'. GPS sync doesn't let you have 2APs deliver their full individual throughput simultaneously, they share the channel and therefore throughput. It's a solution to mitigate interference, not create more throughput.
 
ste
Forum Guru
Forum Guru
Posts: 1815
Joined: Sun Feb 13, 2005 11:21 pm

Re: gps sync

Mon Sep 15, 2014 9:11 am

No. You can use same channels back2back. The APs do not interfere due to GPS and the CPEs reach the second AP from the back side.
The AP Antennas need a good Front2back ratio.
I said 'at full capacity'. GPS sync doesn't let you have 2APs deliver their full individual throughput simultaneously, they share the channel and therefore throughput. It's a solution to mitigate interference, not create more throughput.
No. Back2Back the APs can receive and send at the same channel at the same time. So you double capacity of the tower. A sending AP do not listen while sending. So it is not disturbed by the second AP sending at the same channel. CPEs do only receive the signal of one of the APs. And when front2back of the AP Antennas is good each AP does only see his CPEs.

Of course this does only work when implemented correctly. This is not the case with UBNT. So dont blame GPS sync for bad engineered gear.
 
syadnom
Member
Member
Topic Author
Posts: 405
Joined: Thu Jan 27, 2011 7:29 am

Re: gps sync

Mon Sep 15, 2014 5:45 pm

No. Back2Back the APs can receive and send at the same channel at the same time. So you double capacity of the tower. A sending AP do not listen while sending. So it is not disturbed by the second AP sending at the same channel. CPEs do only receive the signal of one of the APs. And when front2back of the AP Antennas is good each AP does only see his CPEs..
I stand very very slightly corrected and only in very specific scenarios. If you have a 130Mbps (for instance) modulation only 2 access points in the same channel, and they are both operating with equal send and recieve throughputs, then you can send 130Mbps from the tower making the pair of APs full duplex-ish.

There is absolutely no point in GPS syncing back to back radios. Front to back isolation is a vastly superior method because you can completely re-use the channel. If you GPS sync, you will reduce throughput on both sides wastefully. ONLY GPS sync same direction/overlapping APs.

But, that's not the real world. Most all data off tower APs is going to be async ratio like 70/30 tx/rx. Because of this, you can only 'gain' 30% (or rather gain tx equal to the rx amount on the channel), and this improvement plummets with a 3rd or 4th AP because you only have 100% of the 130Mbps to work with among all APs in channel anyway.

Don't get me wrong, there are gains for GPS sync:

improved compute capabilities (more cpus), port speeds (if stuck at 100Mbps like a rocket M5), and ability to run wider channels so high speed plans are possible. Anyone that uses ubnt rocket M5s will probably agree that about 30 active clients is an ideal peak. If the M5-GPS worked, you could have 60 active clients with 2APs, or 90 with 3APs. Maybe 30% gain in overall throughput because you make a group of APs more 'full duplex' like, but the real gain is more active clients. This is worked around today w/ 10Mhz channels, which is ok but leaves less throughput on the AP so speed plans have to be adjusted down accordingly and channels can't be shoved right up against each other so there is some spectrum waste too.

I would much rather see on-the-wire sync with a dedicated port on the device to sync radios. GPS is added expense and we have the technology to make a high precision clock for a reasonable price now.
 
ste
Forum Guru
Forum Guru
Posts: 1815
Joined: Sun Feb 13, 2005 11:21 pm

Re: gps sync

Mon Sep 15, 2014 11:46 pm

No. Back2Back the APs can receive and send at the same channel at the same time. So you double capacity of the tower. A sending AP do not listen while sending. So it is not disturbed by the second AP sending at the same channel. CPEs do only receive the signal of one of the APs. And when front2back of the AP Antennas is good each AP does only see his CPEs..
I stand very very slightly corrected and only in very specific scenarios. If you have a 130Mbps (for instance) modulation only 2 access points in the same channel, and they are both operating with equal send and recieve throughputs, then you can send 130Mbps from the tower making the pair of APs full duplex-ish.

There is absolutely no point in GPS syncing back to back radios. Front to back isolation is a vastly superior method because you can completely re-use the channel.
If you completely isolate the two APs. This is very difficult to achieve. You need a *lot* of shielding. And there should be no reflections.
If you GPS sync, you will reduce throughput on both sides wastefully. ONLY GPS sync same direction/overlapping APs.
No. You dont loose throughput with GPS Sync. You are less flexible as you need fixed rx/tx timeslice on all synced APs. So in some scenarios you waste throughput if usage pattern does not match RX/TX Rate.
But, that's not the real world. Most all data off tower APs is going to be async ratio like 70/30 tx/rx. Because of this, you can only 'gain' 30% (or rather gain tx equal to the rx amount on the channel), and this improvement plummets with a 3rd or 4th AP because you only have 100% of the 130Mbps to work with among all APs in channel anyway.
You have to select a 70:30 Sync rate with this usage pattern.

Don't get me wrong, there are gains for GPS sync:

improved compute capabilities (more cpus), port speeds (if stuck at 100Mbps like a rocket M5), and ability to run wider channels so high speed plans are possible. Anyone that uses ubnt rocket M5s will probably agree that about 30 active clients is an ideal peak. If the M5-GPS worked, you could have 60 active clients with 2APs, or 90 with 3APs. Maybe 30% gain in overall throughput because you make a group of APs more 'full duplex' like, but the real gain is more active clients. This is worked around today w/ 10Mhz channels, which is ok but leaves less throughput on the AP so speed plans have to be adjusted down accordingly and channels can't be shoved right up against each other so there is some spectrum waste too.

I would much rather see on-the-wire sync with a dedicated port on the device to sync radios. GPS is added expense and we have the technology to make a high precision clock for a reasonable price now.
GPS is not that expensive. Look at the ePMP.
 
syadnom
Member
Member
Topic Author
Posts: 405
Joined: Thu Jan 27, 2011 7:29 am

Re: gps sync

Tue Sep 16, 2014 12:15 am

GPS is not that expensive. Look at the ePMP.
debatable. Cambium had a LOT of Motorola expertise at hand so didn't have to scratch start a GPS sync system, saving a ton of money. No, the tech isn't terribly expensive, but the epmp pricing is significant enough over ubiquiti or mikrotik to keep people from using it en-mass for greenfield deployments, and GPS in the radios is a considerable factor. This is especially true if you measure EIRP on a NanoBeam vs an ePMP1000. Ubiquiti wins by a mile here with a PowerBeam/NanoBridge at 25dBm+~23dBi. Mikrotik is right in the middle. The SXT sits right about where the NanoBeams (compact models) fall as far as gain and output, but the price is right in the sweat spot (like the ubnt models).

Ubiquiti failed to get GPS working on an 802.11 radio, and statements from Robert were basically that 802.11 is immensely hard to get GPS working on. The Cambium folks did manage to get it going, but again, with Motorola's playbook in hand and over a decade of experience w/ GPS sync to pull from.

So price isn't really a matter of unit-cost, but how many dollars go into the engineering. Ubiquiti and Mikrotik would have to pour a lot of cash into a product for real GPS sync vs Cambium who had it on the shelf. Much less might be spend with a hard-wired sync cable and good high precision clocks which are available now. 5-10 years ago a clock with enough precision was simply too expensive and using the very high precision GPS satellite clocks made sense.
 
Zorro
Long time Member
Long time Member
Posts: 676
Joined: Wed Apr 16, 2014 2:43 pm

Re: gps sync

Fri Sep 19, 2014 5:41 pm

really working such stuff(about coherence and co-existence within same space, RF-space and time)become when some companies/chipmakers release/widespread cheap/simple to use/integrate chips/devices for that.
recent Quallcomm chips, IP/patents and adds - look relatively promising start in such respect/area.

p.s.
having Navstar portions and functionality in your devices can/would lead to dramatic/critical issues, sometimes/frequent.
so despite relying something from space, beside how it SOUNDS and looks good and simple, Indian or Chinese companies went to IEEE 1588v2 route. some asian companies invented in-house bogus/sub-standard counterparts/clone of such PTP.
 
ilnicchio
Frequent Visitor
Frequent Visitor
Posts: 50
Joined: Sun May 28, 2006 6:11 pm

Re: gps sync

Thu Sep 25, 2014 5:16 pm

We cannot wait to long to understand what mikrotik will do about wireless performance and stability.
The better tech it's now at the same price.
So we need to understand what's the roadmap because without this we need to migrate to better performing tech.
Considering this I can say that ubiquity has worked on RF filter capacity, cambium on GPS... and Mikrotik what's doing?
We are mikrotik WISP from 2005... please keep us competitive also today!
Mikrotik! Help us! :)

Who is online

Users browsing this forum: No registered users and 59 guests