what exactly does selecting "back" or "front" do in a GPS sync'd ABAB frequency reuse scenario

Yep, that is the page.  I hear what you're saying, but the footnote on that page reinforces my belief that the author did intend to write what he/she wrote.  There are 3rd-party 60 and 45-degree antennas that do meet the footnote's side lobe spec, and the OEM 90-degree sector would work fine in an ABCABC setup with no footnote needed.

My challenge is that we are trying to use ePMP in an extremely noisy urban RF environment.  The OEM 90-degree sector is just too wide and sees to much.  If we use an antenna with a narrower beamwidth, we get acceptable performance.  Even with the narrower antennas, it's hard to find two 20 MHz channels, yet alone three, on which to operate.  Being able to use an ABABAB or even an ABABABAB configuration would give us the added flexibility we really need.

Kevin

1 Like

ababab isn't going to happen,  ABAB can.   you need 32 DB CIR, i've never seen any sector antennas drop 20 db in 60 degrees let alone 32.      if you are urban, 5ghz will likely suite you better.     if your more  rural, ABC in 10mhz or 20 mhz channels  on 2.4 work well with the recommended antennas (i actually just posted photos of this arrangement in the "your stores" segment.  

 also, the EPMP system only supports front and back antennas (it has a special mechanism that needs enable for ABAB  or ABCABC to work.  so 3 antennas using the same signal won't align properly. it may work enough to kinda go, but won't be right. 

If you  do manage to find antennas that drop 32 DB in just 60 degrees and has a 60 degree wide face side, I'm sure we'd all love to have them, me especially.   

if you can't find them, good sector tilt will go along way, we limit our outer edges to 3 or 3.5 miles with tilt, it significantly reduces the noise they hear, it doesn't completely eliminate the noise, but helps a lot in town. 

2 Likes

Hi Kevin;

Maybe the new RF Elements Symmetrical Sector antenna (scalar horn antennas) may be something to consider and look at.  They even did a webinar on it earlier this year.  So it can give you a pretty good idea on how their new antenna technology works.

http://rfelements.com/products/wireless-broadband/symmetrical-horn-antennas/overview/

Regards.

Here is a question I have...  If all sectors have internal GPS and are firing at th same time why are the front / back colo options necessary any way?  I do not see the difference having front / back enabled for two sectors versus just having them synced without utilizing the front / back settings.  We usually have ABCD to work with any way, but I would still like to understand the differece front/back make in the sync process vs. just having them gps timed for TX between the APs.  Thanks


carullos wrote:  ...but I would still like to understand the differece front/back make in the sync process vs. just having them gps timed for TX between the APs...

Hi Carullos.  I think Cambium considers this as part of their ''secret magic sauce'' and I don't think they are going to reveal too many details about the actual how/why of this particular setting.  They have previsouly said that this is the way that one AP knows that another AP is trying to share the frequency. But they've also said that the exact technique of what the mechanism tweaked is ''why ePMP works'' and why others who have tried to do that have failed.  So - Im guessing that they are not going to reveal much more of their 'intellectual property' other than - that's how each AP knows another AP is there.  :)

1 Like

A higher level conversation on this list is not going to lead another manufacturer to fix their products or steal the secret sauce.  I simply want to know it I set app ePMP sectors on different freqs on one tower to the same uplink and downlink settings, distance etc without using the ront and back settings (read - like PMP450) do they all TX att eh same time?  A simple yes or no would be fine.  I assume yes because thats how everything else works and why the frame settings have to be the same between APs.  Assuming this is the case, I can't imagine what front / back setttings do.  Don't really care so long as they all fire at the same time...  but if they are not firing at the same time and require front / back settings to do this even when on different frequencies then I believe there is a gross misunderstanding and expectation about how sync benefits between towers and sectors.  Thanks


@carullos wrote:

A higher level conversation on this list is not going to lead another manufacturer to fix their products or steal the secret sauce.  I simply want to know it I set app ePMP sectors on different freqs on one tower to the same uplink and downlink settings, distance etc without using the ront and back settings (read - like PMP450) do they all TX att eh same time?  A simple yes or no would be fine.  I assume yes because thats how everything else works and why the frame settings have to be the same between APs.  Assuming this is the case, I can't imagine what front / back setttings do.  Don't really care so long as they all fire at the same time...  but if they are not firing at the same time and require front / back settings to do this even when on different frequencies then I believe there is a gross misunderstanding and expectation about how sync benefits between towers and sectors.  Thanks


Hi carullos, 

If the APs have the same DL/UL ratio and have the frame size, the will start and stop TXing at the same time. In ePMP, unlike PMP 450, the Max range (distance) settings between the APs do not matter and does not impact frame start times.

The "Front Sector" and "Back Sector" setting is only required if you plan to reuse the same frequency on back-to-back sector on a tower or sectors on two nearby towers that can hear each other. 

Hope that helps. 

Thanks,

Sriram


@carullos wrote:

A higher level conversation on this list is not going to lead another manufacturer to fix their products or steal the secret sauce.  I simply want to know it I set app ePMP sectors on different freqs on one tower to the same uplink and downlink settings, distance etc without using the ront and back settings (read - like PMP450) do they all TX att eh same time?  A simple yes or no would be fine.  I assume yes because thats how everything else works and why the frame settings have to be the same between APs.  Assuming this is the case, I can't imagine what front / back setttings do.  Don't really care so long as they all fire at the same time...  but if they are not firing at the same time and require front / back settings to do this even when on different frequencies then I believe there is a gross misunderstanding and expectation about how sync benefits between towers and sectors.  Thanks


but why give anyone else a clue?      they've done what essentially everyone in the field said can't be done with a low cost PHY.... I wouldn't even begin to open my mouth about what mechanisms have been put into play.     what it does is  everything that has been over looked by everyone else.         The cambium team has shared more than i ever thought they would about under the hood of the EPMP radio, i completely understand why they wont explain the flux capacitor to anyone, even at a high level.  ;)

1 Like

Sriram,

I assume you would know the answer but...I have absolutely seen the distance setting affect performace on synced APs, also from the  optimization guide... This parameter is set based on the farthest SM being deployed. Calculate the distance of farthest subscriber module using Link Planner or any other suitable tools. Set this value 1 to 2 miles higher than the calculated number. Do not set this parameter to the max value as it will waste time allocated for data transmission when it is not needed

it will waste time for data transmision...this would indicate to me  that changing that value would affect the timing frame, if I am wrong just disregard and I will go on doing it as I have always been since it works. If I am right I just wanted to make sure people are not shooting themselves in the foot.

-Ben


@Ben Backus wrote:

Sriram,

I assume you would know the answer but...I have absolutely seen the distance setting affect performace on synced APs, also from the  optimization guide... This parameter is set based on the farthest SM being deployed. Calculate the distance of farthest subscriber module using Link Planner or any other suitable tools. Set this value 1 to 2 miles higher than the calculated number. Do not set this parameter to the max value as it will waste time allocated for data transmission when it is not needed

it will waste time for data transmision...this would indicate to me  that changing that value would affect the timing frame, if I am wrong just disregard and I will go on doing it as I have always been since it works. If I am right I just wanted to make sure people are not shooting themselves in the foot.

-Ben


Hi Ben, 

 

You are correct in that the distance settings affects performance but it does not affect the ability to sync two APs. This is because the calculations required to determine downlink and uplink framing are designed such that transmissions will not overlap. In other words, once you set the frame size (2.5 ms or 5 ms) and the Duty Cycle on two APs the same, the frame timing (Tx portion and RX portion) are the same between them. The distance settings govern how much time needs to be "wasted" to accomodate the propogation delay from AP to the SM and vice versa WITHIN the Tx portion of the frame or WITHIN the RX portion of the frame. The actual Tx start and stop does not change within the full TDD frame. Therefore, uneccessarilly setting the Max Range too high will result in increased "wasted" time (to accomodate calculated progpogation delay) which means less time to transfer real data but it does not impact the ability to sync two APs. This is why we recommend setting the Max Range just above the distance at which the farthest SM is deployed in the sector. 

 

Hope that helps. 

 

Thanks,

Sriram

1 Like

Is there a table out there that relates expected bandwidth vs duty cycle and distance? It would be nice to have a rough idea for site planning. We have settled on 9miles as it seems to provide the best bandiwdth and distrance from our emperical evidence, but if we can increase our outbound by x amount by building more sites to cover the same area it would be nice to know.


@Ben Backus wrote:

Is there a table out there that relates expected bandwidth vs duty cycle and distance? It would be nice to have a rough idea for site planning. We have settled on 9miles as it seems to provide the best bandiwdth and distrance from our emperical evidence, but if we can increase our outbound by x amount by building more sites to cover the same area it would be nice to know.


Hi Ben, 

There isnt a table that relates the two parameters since there are more parameters that impact the bandwidth you can expect. Also each deployment is unique in terms of number of SMs, distance of each SM, MCS levels achieved on downlink and uplink per SM, terrain etc.  I recommend plugging in values into the ePMP Capacity Planner or even better use LINKPlanner to predict the performance. In fact, impact of Max Range to sector performance is negligible compared to the other parameters I listed above. 

Thanks,

Sriram

the speed penalty from propagation delay is related to the speed of light,   at 40 miles, your looking at a little less than 3% penalty, if I did my math correctly. as Sriram pointed out, the propagation delay is the least of your worries at a range were it will become on an issue.   speed of light is 186 miles per millisecond, 186,000 miles per second.  

at 40 miles, your looking at a little less than 3%

Thats what I was  wondering we installed pretty much everything at double 15, it sounds like worrying about keeping your distance setting fine  tuned would be a waste of time.

Sorry to dig up an old thread but rather than explainin what "Front sector" / "Back sector" actualy does can someone explain hwo to choose which sector should be front and which sector should be back in an ABAB setup please?

N

AFAIK, you just arbitrarily pick that in a pair of your sectors.

So, for us, we use ‘North, South, East, West’ for ordering in our minds… so we consider North to be “Front” and South to be “Back” in that pair… and then do the same for East/West as Front/Back in their pair.

That way, all North sectors in our Network are all set to “front,” and any South sectors on our Network are all set to “back”, and therefore we figure that all “fronts” should isolated from each other by either channel separation, or filtering, or shielding, or distance.

the front to back doesn't have to match other towers unless you reuse the channel. 

any "front" ap on channel 5835 can hear any number of back APs on the same 5835, but it cannot hear another "front" louder than -90 otherwise, you will take interference... I don't know what is happening under the hood that causes this, but you will need to watch for that.   


@Chris_Bay wrote:

the front to back doesn't have to match other towers unless you reuse the channel. 

any "front" ap on channel 5835 can hear any number of back APs on the same 5835, but it cannot hear another "front" louder than -90 otherwise, you will take interference... I don't know what is happening under the hood that causes this, but you will need to watch for that.   


Is it other AP's that matter or the subscriber radios from other APs that matter ?  Maybe I'm not understanding the question/answer here but if the AP's all TX/RX at the same time regardless of front/back secret sauce then you should be able to set every single AP on every tower to the exact same frequency and they would never see each other. Of course this would render your network unusable because the Subscriber Radios that can see more than one AP or be seen by more than one AP would not work at all  because they can't see their own AP due to Interference from all the other APs or because their AP is being over powered from interference from the SM's on other AP's interfering with it.  

If AP's are hearing other APs' then they are not in Sync. The entire point of GPS is to keep AP's from interfering with each other leaving only the SM's connected to different APs as a source / recipient of interference.

This whole Front/Back secret sauce thing concerns me because I don't understand why I need it on ePMP when I don't need it on the old Canopy or the 450 gear. It's my understanding that all on the old canopy and 450 stuff all the GPS does is make the APs TX and RX at the same time and that's all I need them to do and that makes them work. So here we have ePMP and they say GPS makes them TX and RX at the same time and that's all you nee... wait, no, you need this front back setting on ePMP but we can't tell you what it does cause it's a secret.

The fact that what the documentation says and what the guys here say are often not in agreement makes it even more concerning.  I just want want the  AP's to RX/TX in sync and all the mumbo jumbo over the "secret" front/back setting makes wonder if that is really happening.


@brubble1 wrote:

@Chris_Bay wrote:

the front to back doesn't have to match other towers unless you reuse the channel. 

any "front" ap on channel 5835 can hear any number of back APs on the same 5835, but it cannot hear another "front" louder than -90 otherwise, you will take interference... I don't know what is happening under the hood that causes this, but you will need to watch for that.   


Is it other AP's that matter or the subscriber radios from other APs that matter ?  Maybe I'm not understanding the question/answer here but if the AP's all TX/RX at the same time regardless of front/back secret sauce then you should be able to set every single AP on every tower to the exact same frequency and they would never see each other. Of course this would render your network unusable because the Subscriber Radios that can see more than one AP or be seen by more than one AP would not work at all  because they can't see their own AP due to Interference from all the other APs or because their AP is being over powered from interference from the SM's on other AP's interfering with it.  

If AP's are hearing other APs' then they are not in Sync. The entire point of GPS is to keep AP's from interfering with each other leaving only the SM's connected to different APs as a source / recipient of interference.

This whole Front/Back secret sauce thing concerns me because I don't understand why I need it on ePMP when I don't need it on the old Canopy or the 450 gear. It's my understanding that all on the old canopy and 450 stuff all the GPS does is make the APs TX and RX at the same time and that's all I need them to do and that makes them work. So here we have ePMP and they say GPS makes them TX and RX at the same time and that's all you nee... wait, no, you need this front back setting on ePMP but we can't tell you what it does cause it's a secret.

The fact that what the documentation says and what the guys here say are often not in agreement makes it even more concerning.  I just want want the  AP's to RX/TX in sync and all the mumbo jumbo over the "secret" front/back setting makes wonder if that is really happening.


its AP to AP.  the GPS front to back is to handle timing from AP to AP.     if an SM hears the same channel from 2 different APs no matter the front to back mark with less than 28 db (i go by 30 for good measure) then you will begin to lose modulation steps.  that is the reason the front to back stats on sector antennas are important.  

for example, if you have 5 towers all 1 mile apart, in a straight line, and you neglect to tilt the sectors, all north and south sectors are on 5835, and your north sectors are all front, and all south sectors are all back, and your sectors are tilted as needed, you'll never run into the 2 fronts or 2 backs hearing each other within the 28db mark of the CPEs uplink power.    

what does front and back do?   cambiums secret.  

what happens when 2 front APs hear each other louder than say 80 db?  your uplink modulations suffer and you'll get random disconnects on weaker CPEs etc. etc.     since the epmp is a modified wifi chipset under the hood, I'm going to guess the contention time portion may be the reason for the front and back mark.  that's an item we cannot define in the EPMP, yet essentially exists in all other radios i've played with in one manner or another, even if it's not user controllable in the gui.  I'll take a guess and say the contention slots are fixed, and not many of them. so cambium sliced them in half, half for front, half for back.   being how wifi handles keepalives, and a few other wifi related necessities, the AP probably speaks during those times so a CPE could be overspoken easily in that time slot. though everything else is GPS synced, some activities under the hood may simply have to be left to contention and that's just the way it is.    I'd love for the front and back marks to go away, I'd squeeze a few more APs up in the same spectrum. 

as far as it is a problem, for us it's not. we stack APs really deep, and usually 4 to a pole.   some of the areas we work, we will have 10 or 12 2.4 APs in just a few square miles up on 100' wooden poles using just 3 10mhz channels. 

the front to back has rarely being a series problem, just something to watch for.   its much less of a worry with 5ghz given the spectrum and antenna characteristics. 

2 Likes

The front to back setting has to do with AP to SM management frame communications.  Let's say AP1 and AP2 are pointed north and south respective.  And let's say SM1 is connected to AP1.  Now let's say SM1 can see AP1 at -40 and AP2 at -70.  As Chris pointed out, this is enough SNR for MCS15 and all should run fine.  And all will run fine as long as AP1 is set to "Front" and AP2 is set to "Back" (or vise versa).  If AP1 and AP2 are both set to the same F/B setting SM1 will randomly disconnect from the AP, along with other erratic behavior, for example reporting its RSSI from AP1 as -70.

To put it another way, if a SM sees (can demodulate the management frames of) another AP on the same channel and with the same F/B setting as the AP it is connected to, it will have issues, regardless of the signal.

The reason why you don't have to do this on PMP gear is becaue of limitations in the wifi chipset ePMP is using.

1 Like