AntennaDeployment

TableOfContents

We are running an urban wireless network, We cover around 600sq miles of Nottinghamshire, abit patchy in places but still growing the links. We have 10 nodes covering this area, a mix of lw pro and OSS, with three backhauls. Not all the nodes mesh, but this is due to terrain, very up and down over here. CPEs are CB3's with antennas from 3dbi to 19dbi directional with long runns of coax, typical CPE distance ranges from 1km to 7km max LOS every CPE connection is LOS, this also goes for the nodes , we will have it no other way, if we can not see the node from a roof top then we will not connect  a CPE, it just asking for trouble!! All the repeater nodes are running a directional antennas to grab the signal from the gateway and an 8dbi omni for the local service. We are using auto power and our wireless cards are netgear ma311 running 32mw MAX. We have one master node running professional sector antennas on the top of a tower 150ftabove ground, and 300ft above the city with the node mounted up the tower. The smaller sites run 8dbi Omnis at roof height coax 5mt max, this gives great saturation. Never have to many problems with this config, the master node up the tower see's over 20 neighbour AP's that are not mesh nodes. We also have a user on the same tower running a cb3 p2p on a 18dbi panel 10ft from our node. You have to understand RF and the way that it propagates, I agree antennas and good site planing make the network, I have worked with RF all my working life 16 years and still I am surprised why some things work and some don't. Most I.T bod's don't understand RF at all, but that's ok because that is what keeps me in work.

asdf
yeah i hear exactly what your saying kenny but the fact remains that i have been told over and over again and read you and others say that having strong rf signals wont work.. i dont have the 13th guy.. i have only two nodes that dont see my gateway and are two hops away and one which is the one im on at my ranch and its 3 hops,, i would like for you to tell me what in the holy hell good signals have to do with routing problems and the mesh losing its gateway and when you tell me that.. and not some BS answer then i will say your right... my signals dont flutuate and havent flutuated 3 db in a year.. i havent changed a radio nor have i changed an antenna.. but at the same time as these other issues have been slowly addressed over the 14 months ive been in this the mesh has become more and more stable.. but anyone who says this thing doest go to crap at times for No reason doesnt pay much attention to his mesh.. btw kenny i didnt read one thing you said about the guy in austin going horizontal polarization to get rid of 20db noise and same channel rf interference which is good sound advice.. instead you want to talk about Texas cowboys that dont know anything about rf when the one you refered to was an extra class ham operator the day he turned 21 and got a 2nd class radio telegraph license the same damn day and i took those tests at an fcc office and not in someones living room the way they do it now.. ham is cool stuff but its a joke compared to doing fcc checks aboard ships.. next time you want to teach me something about rf go take a class first.. i dont know anything about networking and i keep my mouth shut on here.. it would serve you well if you did the same about rf.... and next time i offer to help someone i would appreciate it if you would address the issue instead of just mouthing off about a bunch of maybes and mights about a persons antenna issue.. i know how well your mesh works kenny... hell you called me on your voip phone and i couldnt make out but every third word you said.. maybe u need to reduce your power some more because with two T1 lines i dont think bandwidth was the issue

I'm not out to toot a horn. I just wanted to point out that he's in an urban area and the things you & I take for granted out here in the boonies don't work at all in town. I consider myself fortunate to have deployed mesh from McLeod, TX (population 150 counting the 'coon dogs) to the Dallas metroplex which now has over 5 million people (not counting the 'coon dogs) and therefore have a little experience behind the suggestions I give.

So don't take it personal because I was probably one of the first guys that took an omni up a tower in hopes that it would work! Yes, I made several trips up several with different amounts of gain and downtilt up a 1350 foot tall tower and did range tests - or at least tried. Looking back on that, I feel pretty stupid for even thinking that something that absurd would work. And I should have known better because I've been a ham radio operator for 13 years now.

It's a common mistake to think early on that you can push the RF because sometimes it actually WILL work. It's not until you repeat it 3 or 4 times that you seem to see it all go to pot when that 13th guy starts downloading - as I said. I just don't want to see these guys fall flat on their faces by making common mistakes up front. I was lucky that a few of the English chaps and Dave Peterson steered me in the right direction early on and now I want to repay the favor... that's all.

And you're right in saying that Jon has done *great* work in pushing development - but I can tell you that properly setup, build 22, 23, 24, and 25 all meshed well for me. Not that I'd ever want to go back to an ad-hoc only mesh with build 22, but it worked.

Dang man - us Louisiana guys don't even have cowboy hats. I wonder if Dave Peterson has a hat or boots or something like that now that he's moved to TX? :)

Outdoor Meshed AP Distance?
damn i didnt mean to get into a pissin contest about who is the Big know it all on Here... most of us that have been around very long already know who you are,,, god forbid anybodyelse knows a damn thing,,, as far as overpowered nodes on towers some friends of mine have the same exact problems i have trying to mesh nodes that are 400 feet apart and have tried running everything from 10mw to 1 watt so that is BS... and the fact is.. this stuff still has issues and most on here dont know their ass from a hole in the ground when it comes to rf theory and application but ive got 30 yrs of experience and know enough when i hear bullshit.... but i do appreciate the effort that is made to work thru those issues by jon and the guys that work behind the scenes to figure out the little tricks to this thing.. i just have better things to do than to blow my own horn on here so i can be the Smart guy and i only offer advice when i know what im talking about.. all i was doing was tryin to be helpful because some people here have helped me with the knowlege they have and their experience... ive learned onething for sure... the people here who sell stuff for the most part seem the people who are the only ones that have the Perfect mesh.. and when you ask them questions about a specific problem you always get that old Well its this and its that and if thats the case why in the hell dont jon just fix it... but after 14 months of its this and its that i know the difference most of the time now... when i hear the same people make real stupid comments about things that i know from my area of expertise that are theoreticaly and practically wrong makes me wonder if the other stuff they say about things i have questions about and really no knowlege about is just as stupid.. btw jon the oss mesh software seems to be gettin better and better thanks its not the wrong time of the month i am just in the mood to tell it like it is for a change and i own horses too. I think your model should work great. On my new, in-town access points, I'm now using 9 dBi omni's with electrical downtilt. I have mixed feelings about the power. Today I must run power at 200mW because I've got a lot of clients with an Engenius bridge sitting on their desktop (indoor - no outdoor antenna). In talking with others about what works best, those with power levels set to "auto" report more mesh stability than when they had power turned all the way up.

Okay, on the subject of urban deployment, what are mesh operators deploying. We have taken the route of 10-12dbi omni's at rooftop level with 200mW cards turned all the way up and the sensitivity turned down. We figure if me make it unpleasant enough for other wireless devices, maybe they will stay on a different channel. We inject DSL into the mesh for every 2 to 3 MeshAP's to localize traffic. What other solutions have people used in urban areas? BTW, we are deploying in the Denver metro area and have seen a lot of the Best Buy take it home and plug it in, wide open wireless solutions on channel 6.

Outdoor Meshed AP Distance?
Don - I think we are really saying the same things, just in different ways, so don't take it wrong. No one's software can overcome the practical limitations of RF. My point is focusing on maintaining a good, solid signal is the key to a solid mesh no matter where you are deploying - rural or urban. Antennas are more important than anything in my opinion. Heck, even using the wrong omni will screw you over. I can't tell you how many times I've seen guys do things "the cowboy way" - after all I live 2 miles from Texas :) (no offense to the guys from Texas!!). It's easy to go out and stick a bunch of overpowered omnis up on some towers a couple of miles apart, but once they get twelve or thirteen people scattered across their network, the mesh falls apart (throughput drops and latency increases to an unbearable level). How many times have we heard people on this list with that complaint? They all say that the SNR looks fine, but they wonder why the traffic just can not get back to the gateway. I believe it is because of all of the contention that is going on. Naturally - everyone blames the software. And not just LocustWorld Pro or OSS, I've heard Q-Code blamed as well. They never consider that it's that "thirteenth guy" in the far corner of the network is pulling down bandwidth and all the other nodes can not pass traffic properly because they all hear each other. Making use of dual-radio boxes and separating the antennas on different channels is the key to making towers or long shots work right. That way one antenna hears fewer clients and fewer access points, and the other interface can relay traffic to the other channel more efficiently. It only costs about $300 (antenna, coax, radio card) to add a second interface to an access point. Spend a little more up front and do it right. Trial and error have taught me a lot. Heck, I wish I had just half of the money I've wasted from errors. However, I disagree with Don - I do think that there is a scientific approach that exists. We have many good models worldwide (both rural and urban) of what works. We also have some good examples of what won't work. I just hate to keep seeing different people keep making the same mistakes. My network now covers three small towns that still have little WiFi outside of what I'm putting out. Still, when I'm installing new nodes today, they either have omnis at rooftop height (and only talk to a neighbor node or two) or a dish on the top of a mast (to talk to a distant neighbor) with an omni on the second interface (for local client and mesh coverage). I now have clients who are connecting to the internet through 10 miles of RF links and they have the same throughput that I have (5 blocks away from the gateway node) and 60ms of latency to anywhere on the internet (6ms of latency to the gateway). I believe the manner in which I try to deploy mesh today will work anywhere. Final story & I'll shut up: I recently made some changes to the access point that Don has a picture of on his blog. It's a dual radio AP and it's the one I mention making about a 10 mile trip (through a repeater node on a tower). The gateway has 2 interfaces, the tower has 2 interfaces, and this AP has 2 interfaces. When I applied the changes, somehow I screwed up the settings on the 2nd interface on the distant node and it went to ad-hoc mode and basically disappeared. The clients connected to the omni interface and it meshed to the tower just fine. The signal looked great. However, they started calling after a few days complaining of dialup like speeds. It was then that I discovered my error and corrected it. This was a 4 mile shot in McLeod, TX, where the population is about 150 people. Imagine what it would have been like if this was in a more densely populated area!

What is the term "there is more than one way to skin a cat." or there are more way to deploy a mesh! The solution to deploying a mesh in a large city is not Kenny's way or Don way or Eddies' or Bishop's or Ted's way. It the way of best practices. Currently there are no best practices to deploy in a city. If there was it would be done by now! Remember trail and error seems to work but for a large city deployment you need need more of a scientific approach. Nortel, Mesh Networks, Bellaire, Wi-lan all have a methodology but they do not have developed best practices that can be tested and duplicated. Look at the Philadelphia mesh. What a mess. To deploy in city takes a lot of research and investigation. You can go half cocked deploy but remember that your mistakes may costly. I caution anyone who want to deploy in a city, research, test, model, verify, ..... do all the things that taught you in school before doing the cowboy way! I never said you couldn't deploy it in a city or any dense urban area... I should have summed it up by saying that you have to factor in all the noise. We had the guys from Dallas come and look at our networks so they could see what was possible, but we told them that they would never be able to pull such off in the middle of Dallas due to the noise. Still, after a month or so, one of the guys called and asked why they weren't getting the ranges we were!

Omni's will work in any deployment as long as you keep one thought in the back of your mind and that is, "besides my other access points, what is this thing going to hear?" If it's up at 180 feet, it will hear a lot... so you'll have to keep your nodes close to one another. Ideally I would use omnis in neighborhoods at rooftop level and keep nodes close to one another. Use sectors on your towers and have one box in the neighborhood with dual radios & a directional on one interface that is linked to the tower. Keep the mesh at rooftop level.

One point that I didn't make is that the choice of proper antennas is 99% of a good mesh. I've gone through more than 2 years of OSS software as well as the Pro software and everything worked well all along because I focused on antennas early on and have good strong links.

aasdf
Hum, Guys, for the last year I have been deploying in a city. Wi-fi is everywhere in the city but to say that you cannot deploy 802.11b through out the city is incorrect. First of all mostly everything in this discuss is way off. Yes you can use Omni's antenna's yes you can intermix with cordless phones, baby monitors, microwave ovens with in the first mile! Even though it's intension is sincere, it sounds too much like a sells job for getting pro version! What is true is that the distance in a urban area is depending on your equipment is really reduced. I call it the urban forest. That is homes, apartments, trees, wall etc. In a rural setting you can easily get 2-3 Km or even better. In a city you hardly can get a single 1 km. It doesn't mater what antenna you use or combination 200mw radio cards and antenna. Your mesh density (the distance between each MeshAP) has to be within .5 KM anything larger than that becomes spotty at best. My comment is not to start a debate on can Locustworld's OS can be deployed in a city. My belief is that it can. However, I have slowly been deploying this mesh in Toronto for over a year and half. I have had many problems that most of the small towns and cities may not have. However understanding the problems and careful deployment you can easily alleviate the problems.

I agree with Kenny, our experience so far in urban areas shows 2.4 GHz over long distance practically useless. The only way we have found to get any distance out of 2.4 GHz in an urban area is with 24db parabolic dish antennas polarized horizontally, in this configuration you can still only expect 1.5 to 2 miles. We have adopted the approach that Austin City Wireless is using for our urban deployments. In their model each access point has its own wired internet connection which are widely available in most urban markets. This totally defeats meshing but we have yet to find a better way to deploy in dense urban areas. We also recommend the shortest feed line possible. Although we do have a couple of runs over 100ft which are working fine for the areas they are in, most of our installations have power run to the top of the tower and into the node, with a very short feed line to the antenna. You are missing something - he's in Austin. You & I have networks in very rural areas. Urban deployments are totally different. Last year we deployed a mesh in the middle of Dallas and it required making a shot to a tower that was about two miles away. It ended up being very difficult and eventually required using 5.3GHz to pull it off (5.8GHz was full!). If he's in a dense urban area, 2.4GHz and omnis are not going to work - period. Those omnis will hear thousands of cordless phones, baby monitors, microwave ovens in the first mile around it, but not his access point over 3 miles away. In urban areas, long-range 2.4 is useless. Omnis are not going to work either.

Bishop - you are spinning your wheels and wasting your money if you throw a bunch of hardware up in Austin several miles apart and expect it to work. You are going to have to link your boxes on 5.3/5.8 using the LocustWorld Pro software clustering option. I would not even use 2.4 to link towers - there is too much interference. True, you loose true mesh from tower to tower, but I would consider installing neighborhood access points in a manner in which a few pull bandwidth from the tower and repeat the signal to other access points that penetrate into areas that the tower can not reach.

Keep your links as strong as possible, and avoid overloading tower access points with too many clients. Remember - if you have a neighborhood access point that has 20 clients on it and it is pulling bandwidth from that tower, the tower sees those 20 clients as one station (the access point). An access point can only have so many clients on it before it becomes overloaded. That is one of the benefits of the mesh - it basically does some load balancing. In an urban setting, you will definitely want to have external CPE on most of the houses, and aim them at the neighborhood access point when possible, not at the tower.

Since your antennas are 180 feet up, I would avoid LMR400 and use LMR600 or higher. Yes, it's costly, but Eddie makes a good point with his 240 foot run. Ideally, I think it is best to run AC up the tower and put your MeshAP right up there next to the antennas. The shorter the coax - the better!

Good luck and keep us posted on your progress. It seems that Austin is one of the hottest markets for LocustWorld mesh these days. I know of several that monitor this list that are deploying down your way.

3.5 miles should be a cakewalk.. go back and check your equipment... 7 miles you might be on the edge.. get those verticals away from the towers at least 4 feet unless your right on top of the tower.. i work cheap so feel free to give me a call and i will take a trip.. or assist you on the phone for the heck of it.. you failed to mention coax length and type.. i have 50 foot of lmr 400 and runnin 200 mw and shooting it 3.5 miles and signal is no trouble with a 9db gain vertical at one end.. if you dont have lmr 400 or better then u need need to get some.. food for thought.. i ran 240 feet of brand new lmr 400 straight up a tower and couldnt find the signal period with a pda at the antenna so dont believe what they say about db loss per hundred feet.

One thing not mentioned - the RF pattern of 15dBi antennas are almost flat and if they have some electrical downtilt the RF is slightly directed downward. With that much gain, the antennas just are not going to see each other's RF and mesh. Even if they are in the same vertical plane, it's not going to happen if there is some downtilt. What I've found works best is using 120 degree sector antennas. You can tweak the angle of the antennas to give client coverage as well as mesh connectivity. I see you guys are down in Austin. It might be of some benefit for you guys to take a journey a few hours north and see some of the older meshes in LA & TX and how we have adapted. Sticking up a bunch of omnis is the "brute-force" approach. It works well, but it's costly. I believe in getting as much out of each access point as is possible. Use of the right antennas is a key factor.

omnis directional deployment
High gain omnis have a very narrow vertical radiation angle. If your tower installations is anything but exactly vertical, the radiated pattern will be either into the sky or into the ground. Use low gain omnis for client access and create multiple "micro cells" (i.e use more MeshAPs). Then use directional antennas for longer links.