900 MHz and nLOS, NLOS, and LOS

Hi Everyone,

I thought I'd share some observations about operation at 900 MHz, and it's ability to deal with trees and such.  I've had a lot of operators try 900 MHz, some with outstanding results, and some...not so much.  I can attibute this to the non-line of sight operation and foliage penetration claims and capabilities of these lower frequencies, and some misunderstandings about what that means.  First - I'll talk about the successes.  Those that are enjoying the benfits of the PMP450i and the throughput enhancements generally have one big thing in common, they are deploying 900 MHz just like they would deploy 2.4, 3.65 and 5 GHz products.  That means from high access point locations.  By high, I genrally see best success at a minimum of 50ft over the treetops.  Like any band, higher can be better.  The exception is if getting higher increases the over the horizon noise floor of the AP.  Those 900 MHz signals can travel, and sometimes that is not such a good thing.  Fiding the the right balance can require multiple spectrum scans on the AP as you try different altitudes.  Sorry - if this was easy, everyone would do it.  All of this ties back to penetration performance.  Things work best if you can get a good "launch distance" of clear air before you hit the trees.  As I often say, 900 can get through some stuff...but it needs a runing start.  Many of the failures I have seen have had both sides down low, and close to the trees.  That gets to the second point.  Get as far back from the first obstruction as you can.  Even if it means going to the far side of a building away from the tower. 

The next place where folks trip up, is not looking before they go on air.  The 900 MHz band is from 902 - 928 MHz here in the US & Canada, and in some places like Austrailia it is only half of that.  This means that all the stuff that wants to take advantage of the license free band with better penetration needs to live in a pretty small slice of spectrum.  Firing up your spectrum analyzer and letting it run for a minimum to 10 minutes is a must, before selecting a channel.  Fortunalty you bought PMP450i gear so when you select a channel you can re-use it back to back.  There is a good chance that in many areas, you won't be able to chose a 20 MHz channel, in fact smaller channels of 5 and 7 MHz are more commonly used.  You'll often get more thoughput through a clean 5 MHz channel than you will a noisy 20 MHz channel.   Chose wisely, and sometimes the best you can do is not as clean as you'd like.  The good news is that in many areas the really noisy 900 MHz paging is largely gone but if you do encounter it the PMP450i has an extra filter setting to help.

If you do have to run a link low and through the trees hopefully it's not too long.  Some informal testing suggest that wider anlge panel antennas can outperform a higher gain yagi in these scenarios.  I'm still gathering info on this but the early tests are encouraging.

I'm sure others have useful observations to contribute.  We'd love to hear them.  Happy Linking.

what you say makes sense as I have been struggling with getting our links to work. But the one thing I guess that needs to be addressed is that Link Planner shows workable links which is what we used to decide where to deploy the AP.  I get that link planner can't compensate for foliage, but this sort of negates our work in link planner it seems.

I find that noise is a huge factor, and also extremely unpredictable.  You can have two sites in a subdivision without 100m of one another, with the same signal, but one has huge noise issues while the other is just fine.  The only way to see if noise is a factor is to go out and try it.