Possibly multipath problems? Tips on resolving?

Products PMP
1

We have a strange issue with some of our customers connecting to a 450i access point.

It is a 90 degree sector serving a total of 16 sm's with distances ranging from 0,3km to 4,2km.  Every subscriber have clear LOS, and signal strengths ranges from -56.5 to -68.5 in the downlink (see attached "LinkStatus" page from the ap for more in depth values).  The sm's are primarily 450i integrated, with one customer (my own holiday cabin actually) having a 450 with clip.

The AP is configured with 75% downlink, range 5km, contention slots 8 (to keep consistent with sync of our other ap's in the area), frame period 5ms, sm receive target -54dBm.

The problems we are having, is that some of our customers starts disconnecting, suddenly having poor SNR, and even sm rx signal strength seems to drop. Only solution to this is to change the frequency, or move the sm at the customer location.  After a while (might be a day, a week, or even months), the problem reappears, and we have to change frequencies yet again.  Spectrum-scans from both the ap's and the sm's show very clean channels.

The "funny" thing is that this happens only to the sm's concentrated in a certain area, all other sm's are working great on every single frequency and bandwith tested. See attached "MapSMs" for layout of all the sm's connecting to the ap, and where the problem-area is (marked by red circle). Do not bother the few offline sm's nearby the ap, as theese are holiday cabins where some customers unplug their sm's while not beeing there.

Every sm not in the marked area works as a charm, with no issues what so ever.

I am wondering if this might be caused by some multipath reflections in the area?  Another strange thing I've seen, is when changing the frequency from for example 5,7GHz to 5,4GHz, resulting in 6dB lower output from the AP, the sm's in the specific "problem-area" does'nt always get 6dB lower receive signal (as I would assume) - sometimes they get far lower than 6dB receive signal (I've seen as much as 13dB) and sometimes even stronger signal than they had before reducing ap output....  This to me, seems like there might be some multipath in the area affecting sm's receive power either positively or negatively.

This is in the Norwegian mountains, which has a good deal of snow in the winter.  It seems as theese problems are worse in the periods of transition between winter and spring and/or when the temperatures fluctuates between above freezing in daytime, and below freezing at night.  I'm suspecting theese changes in temperature might affect possible multipath issues, and might be an explanation to our problems, but I would appreciate some feedback from others if you have any explanation or tips!?

If this in fact are multipath issues - how to best deal with it?

As mentioned - theese are 450i sm's, which has about 10degree azimuth/elevation, so changing to a narrower beam at the sm, means we would have to use 450b-High-gain which has only 7degree azimuth/elevation.  The big issue with this, is that the mountains is VERY windy, so the wind-loading of the 450b's would be a big issue (to maintain LOS, some of the antennas are obviously mounted above the roof), as well as that the area is restricted regarding sizes of allowed antennas - parabolic antennas are not allowed...

The other possible solution I come up with, is trying to change the ap antenna to a narrow-beam one, hoping the signals will not "hit" the reflection point causing the potential multipath, but honestly I do not think it'll change anything (as I assume the reflections is quite close to the customers).

The last thing I can think of as a solution, is to try a 450m as AP instead of the 450i, as I think the 450m has both beamforming and beemsteering(?), possibly resulting in less multipth-reflections.

Theese are my theories, but I am certain that someone in here are able to help with other possible theories and/or tips on whats best bet on resolving this issue!? 

1 Like
2

We are in sunny FL , USA and are now experiencing similar behavior on both clients across a range of 450SM clustered in a small area being fed by 450M & 450i APs.  

Was this issue resolved?  If so, would you care to share?  Thx.

3

The issue has not been resolved - it seem like this time of year the issue is'nt as much a problem as it is in the winter.  It is quite difficult to troubleshoot while the problem is'nt showing, and I have not gotten any replies to my post, so at this point I have no tips for you I'm afraid.

I was hoping some Cambium-staff would reply at some point, but they have'nt...

4

If the problem is a multipath reflection 180 degrees out of phase and cancelling the signal, you will see it affect the signal in both directions, i.e. at both the AP and SM.  A signal graph will often show a pattern, like a fade at sundown or sunup, or on windless days.  If you use tone alignment, sometimes you can hear a warbling not steady tone.  We find the problem tends to be more common when shooting across a very shallow valley or dip in the terrain.

I think you are unlikely to solve it with narrower antennas, although more antenna gain can give you more margin against fades.  Changing frequencies is also unlikely to be more than a temporary fix, unless you actually change to a different frequency band, that still may not help.

We try moving the SM up or down a couple feet, tone alignment can help find the best spot.  In our case we have growing corn or soybeans in the fields, so this may only solve the problem for a few weeks until the crops grow.  Counter intuitively, sometimes going lower works better than going higher.  It's more an art than a science.

The only sure fix I know is to mount the SM where it shoots over a building or some trees, so it only sees the main path, not the reflection.  Trying to do this with a narrower antenna beam is usually not feasible, you would need a beamwidth of something like 1 degree.

1 Like
5

Thanks for the reply!

As you mention, adjusting the sm's higher or lower seems to have some effect - the problem is that this issue "comes and goes", so when adjusting the sm's higher it's ok for a short (or longer) period, and then suddenly the sm's have problems once more.  Then adjusting the sm's back to where it was initially, might solve it once more.  Adjusting all the sm's time after time is'nt really an option, as we would have to have techs on standby in the area at any given time, and the customers can't be expected to move the antenna themselves every time this happens. The sm's are all shooting above other buildings and/or trees - I suspect the issue might be because of reflection by the snow (this is in the mountains) when the temperature rises above and below freezing, so the snow surface sometimes are like a "mirror", and sometimes are very moist, and other times very dry.  Theese issues does'nt seem to be there in summertime when there's no snow...

I am really frutrated, as our customers of course experience serious problems from time to time, and we do not seem to be able to make a permanently good connection.  We have aproximatly 330 customers in this mountain-area, with only this small area (of about 10 customers receiving from this one AP) having problems. Even other customers receiving from this AP, but are in another area are working very well...

We allways use alignment tone when adjusting by the way - and then confirming the signalstrenghts comparing it to expected from linkplanner...

6

If the problem customers are all in one area, maybe you could set up a microPOP site for them?

Otherwise, the problem might not be solvable, and you would need to just let those customers go, unless they are willing to live with the intermittent service and use something like a mobile hotspot as backup.

If LEO satellite Internet like what SpaceX is doing works out well, we might be able to worry less about a small percentage of unreachable customers.  For us, it's the people who build their houses in mini forests.  Of course, if Starlink works out TOO well, it could put us all out of business.

7

If it is multipath, and it sounds like it is, one solution that has served me well in the past with PTP links is to orient the antennas up so the link loses 1-2 dB from its peak. This is more effective when using high gain antennas. What this does is it puts the reflection farther off of your main lobe, usually the multipath is reduced more than the 1-2dB you lose from your main path. This is most effective on PTP links when you change the orientation on both ends of the link as it has a cumulative effect vs just the one end in this case. You will want to avoid moving it down 3dB as you will then be dangerously close to the edge of your main lobe and that will cause deep RSL fades.

Often times you see multipath like this when going over water or very flat terrain with little vegetation or buildings in the path to obstruct the multipath. You can use LinkPlanner to adjust the height of the antenna so the Fresnel zone is just clearing the ground werever they meet. This will give you the best overall protection from multipath but it doesn’t guarantee it will never happen.

8

Thanks for your reply ericw!

A strange thing is that there are both buildings and vegetation spread out in this particular area, so the different customers all should have quite different multipaths I would assume. It is not going over water, and the terrain is not flat, but crossing diagonally over sort of a valley (attached is the link-profile from one of the sm’s, where I put a red dot where there are many other buildings, and green where it’s forest). Changing the tilt seems like a possibility of course, but when changing frequency, the issue is gone for any given time before it reappears - I would imagine it might be the same if adjusting tilt, as the possible multipath seem to change at any given time - seems it might have a connection with snow melting and freezing again).

image
image1629×299 71.1 KB

9

This has every likelihood of being multipath. Facts you have given which support that are a) that 8 out of 18 SMs on one sector have this problem, b) all of the problem devices are nearly the same range and angle from the AP, and c) the 10 devices in different areas do not have any problem. The snow is not normally very reflective but when the surface turns to water it becomes very reflective, water having a dielectric constant of 80 this surface does not have to be very thick. The shape of the surface of the snow can be such as to form a near parabolic reflection enlarging the signal relative to the direct path.

Given the path profile it looks as though the likely location for the reflection is about 170m from the AP. In this case there are a number of possible solutions. It does not appear to be caused by the usual problem of a lake. The spacing of the affected devices suggests that the problem is not at the SM end of the links.

  1. Use some sort of obstruction to screen the antenna from the ground reflection at 100-200m from the tower in the direction of the affected subscribers. The obstruction would need to be about 3m height (more if the snow regularly gets deeper than 3m) by 8m width (to cover the subscriber directions). The obstruction needs to be reflective such as metal. It could be a mound of earth which would break up the reflection.

  2. Raise the height of the 450i by about 15m so that the reflection is in a part of the gain characteristic which is lower. This may not be very effective with the integrated antenna because the antenna employs null fill in the lower portion of the characteristic to ensure that nearby SMs below the beam receive good signal as well as distant SMs in the beam.

  3. Tilt the antenna up about 2° so that the gain to the reflection point is lower. This method is not recommended for the same reason as 2 above. It would also not be good to sacrifice gain to nearly all of the SMs.

  4. Invest in vertical space diversity. Take one Connectorised PMP450i and two Connectorised Sector antennas. Connect one antenna as Vertical and the other antenna as Horizontal, put 50 ohm loads on the other two antenna ports. Place one antenna at the current height and the other antenna 45cm higher or lower than the first antenna. When the signal cancels at one antenna it will be a peak at the other antenna for this geometry. When a null forms on one of these antennas the throughput will reduce to approximately half using MIMO-A, but will not drop to zero.

Technique 4 is particularly effective when a lake is involved. It is used quite often with PTP600/650/670 links. LINKPlanner assists with the design process by recommending the vertical spacing of the antennas.

I hope that one of these techniques can be used to solve your problem. Technique 1 is usually the simplest to implement, particularly if you can change the shape of the ground at or near the reflection point.

2 Likes
10

Wow - thanks for the great reply Nigel!

Here are som good tips for us moving closer to a solution (hopefully).
I don’t think solution nr. 1 is that easy - changing topology is probably not very well welcomed by the the cabin-owners in the area. Regarding raising height as in solution nr.2, our tower is just maybe 1 or 2 meters heigher than where the ap sits now, so we are not able to do that unfortunatly.
The tilting in solution nr. 3 might be good enough - I’ll have to check the linkplanner to see what effects it will give in total.
Solution nr.4 would maybe be an option, but then I have a couple of questions if any of these alternatives would give us similar results:
1: Changing the ap to a 450m - would that help because of beamsteering-functionality?
2: Would it be an idea to put up an extra 450i AP with more directional antenna - just focusing the beam to the “problem-area”? With like a 10degree Mars antenna (like in linkplanner)?

Appreciate your response!

11

I did some further analysis with the LINKPlanner file. The problem SMs are in the very narrow ranges below.

  • Bearing range 341.6° to 343.5°
  • Path Length 3.897km to 4.129km
  • Path Declination 1.3° to 1.9°
  • All SMs Declination 11.4° to 1.3°
  • The reflection is approximately 170m at 7° Declination.

Changing the AP to 450m would not help since the device has no vertical diversity.

Adding another 450i and putting a dual polar 2ft dish or flat panel would be very helpful. The beamwidth of these is about 5° which easily accommodates the wanted problem SMs. The attenuation at 5.5° below the beam should be sufficient to attenuate the reflection by 6dB or more which will ensure that the reflection cannot form a null.

A photograph of the base station from about 250m at bearing 342° would be very helpful to ensure that my analysis is correct. Google earth is very low resolution here.

12

I’ll see if I can get a photo as requested - we might not be able to get a photo of the base station at 250m as it seems like we have to be at least 5 or 6 meters above ground to get line of sight, and there are no buildings or anything to climb on top of at 342degrees. We can take some more pictures at different distances in the same bearing of course, given that we have line of sight to the tower.

As you mention - a 2ft dish or flat panel with 5degrees, do you have any recommendations to which types are a good choice (linkplanner only has the 10degree mars)?

13

As far as the photo is concerned, I simply want to see the ground and its relationship to the base station at the 170m point at 342°.

Cambium Networks 2ft Dual-Polar Parabolic RDH4503C (29.5dBi) is the recommended antenna. It is listed for Point to Point links only because it would not normally be required for Point to Multipoint. (The high performance dish is only listed for requirements where good front to back performance is required for frequency reuse reasons in PTP applications.)

14

Ok - thanks. I will send a guy taking photos tomorrow (hopefully), so I’ll provide you with them once I have them.

15

I’ve had a guy taking some photos today. He took quite a few of them, and here they are (compiled into a word document to ease the explanations for each picture) - let me know if you would need some of the photos in other format:CompilationPhotosProblemAP_Haglebu_for_Cambium.docx (848.9 KB)

16

Hello Rambern,

I have encountered such an issue where SMs disconnect randomly. Usually I have seen on top hills with a good line of sight to client SMs as it is for you. I am based in Rwanda. Though, we usually reboot the PMP 450i AP and the issue is sorted. Though, some of these places are usually close to RADAR system equipment for air navigation.
We have deployed the PMP 450m AP to sort this at some locations and it worked due to the AP high gain/beamforming. We have also just changed a bit the location at the same height, but different leg of one PMP 450i AP on the same time and changed frequencies and the issue was fixed. Try also to use smaller channed width (15 MHz if this can accomodate client throughput) and see the result.
Also

  1. did you scan with the integrated spectrum analyser from the PMP 450i AP and problematic SMs? If not yet, do it. Make sure, the channels used are clean.
  2. Have also a SM received target level at least -55 dBm configured on the PMP 450i AP.
  3. How much transmitted power do you use on your PMP 450i AP?
  4. Do you use GPS sync to avoid self interference in your network on all your towers.?

Let me know your oppinion.

Sincerely yours,

Niragira Olympe

17

Hi Rambern,
The pictures are excellent but I realise that I should have asked for them closer to the tower. Initially I was concerned about the surface of the road, but I think the road is reasonably well screened from the radios on the tower. Consequently I believe that the reflection must be coming from the ground around 170m.

You might like to walk the road and check that it does or doesn’t have a good view of the tower ie. with or nearly with Fresnel clearance. It clearly has a good view to the subscribers and although the line may not be at 342°, there is a slight left slope on the road which my cause the reflection to be just right. There is also a small curve on the slope of the road which could amplify the reflection.

If the road was the cause, then an obstruction would help or lowering the sector antenna by 4-5m would remove the reflection. Consideration of the consequences to the good SMs needs to be contemplated though and some of them seem to need the height.

18

Thanks Olympe!

We have tried both different frequencies as well as bandwiths, but to no avail.

  1. Yes - spectrum is clean
  2. We have SM receive target set to -54
  3. Our transmit power (eirp) is 36dBm when in 5,8GHz, and 30dBm in 5,4GHz (we have tried both bands)
  4. We do use GPS sync on all our towers
19

Thanks once again Nigel!

Some of the road do have good view of the tower (from my recollection), and some of it not.
Lowering the sector seems to have to much of a negative impact for the rest of the SM’s, so it seems like the best way to go is to add another 450i with a directional antenna I think. Does the directional antenna need to be 5degrees, or could we go 10degrees as well you think? We are thinking about using RF elements Ultrahorn (UH-CC-5-24), as the RF elements antennas has good reviews as well as the availability from our supplier is quite good. What do you think? We would like to “do this right” to eliminate the need to experiment back and forth.

20

Hello Rambern,
It seems the installation is fine. You can try with the Horn antenna or go for the PMP 450m AP. Though, we need to understand the issue where it is happening. Make sure also that the power is stable at the tower (Power injector or CMM4/CMM5) depending on what you are using.

  1. Would you precise the power source and power injector used?
  2. Make sure also there is no RADAR.

Sincerely yours,

Niragira Olympe