Point-to-multipoint basic questions

number9 · 09-15-2004

Hi folks,

I'm a little new to the whole WiFi world, but for a bit of fun I'm hoping to roll out some gear to do either multiple point-to-point connections or a single point-to-multipoint install.

Essentially, I want to connect three LANs (actually, three houses) to our office building LAN (which is one building) in a city. The houses are at most ~3km away from each other, and line-of-sight shouldn't be a problem.

I'm wondering, though, how well this is going to work? The reason I ask is that I've heard horror stories about point-to-multipoint and something called, I think, the "hidden node" effect where packets are randomly dropped. Is this true?

Bear in mind I have a maximum upstream (i.e. the connection to my ISP) of 128Kb/s, I'm hardly going to be doing high-speed connectivity. I just want to be able to browse web & check my email from home; and given that I live in a developing country this isn't possible with traditional internet access.

Any info would be greatly appreciated!

Thanks,

Chris
Bamako, Mali (west Africa)

Thorn · 09-15-2004

Quote:

Originally Posted by number9

Hi folks,

I'm a little new to the whole WiFi world, but for a bit of fun I'm hoping to roll out some gear to do either multiple point-to-point connections or a single point-to-multipoint install.

Essentially, I want to connect three LANs (actually, three houses) to our office building LAN (which is one building) in a city. The houses are at most ~3km away from each other, and line-of-sight shouldn't be a problem.

I'm wondering, though, how well this is going to work? The reason I ask is that I've heard horror stories about point-to-multipoint and something called, I think, the "hidden node" effect where packets are randomly dropped. Is this true?

Bear in mind I have a maximum upstream (i.e. the connection to my ISP) of 128Kb/s, I'm hardly going to be doing high-speed connectivity. I just want to be able to browse web & check my email from home; and given that I live in a developing country this isn't possible with traditional internet access.

Any info would be greatly appreciated!

Thanks,

Chris
Bamako, Mali (west Africa)

Search this site as well as Google for "WISP" (Wireless Internet Service Provider), as this is essentially what you are attempting to create. While many WISPs are much larger, the difference between what you want to do, and what they do is just a matter of scale.

The Hidden Node problem is this:
A central Access Point communicates with Stations 1 and 2. The AP can see and hear both 1 and 2. Station 1 can only see the AP and not Station 2. Station 2 can only see the AP and not Station 1. The nodes are hidden from each other, hence the name. Since they cannot hear each other's transmissions, it is possible that they will both transmit at the same time and neither transmission will get through.

In practice, it isn't too much of a problem, unless there is a high traffic volume. It can usually be adjusted via the RTS/CTS thresholds on lower volume systems. (Look in the equipment manual for details, or Google.) There are specific hardware and firmware combination to defeat this is it is a problem on high traffic systems, such as ORiNOCO Access Points running Karnet firmware.

number9 · 09-15-2004

Thorn: Thanks for that. As our upstream bandwidth is limited to 128Kb/s, we certainly won't be doing much traffic. There will probably be a nightly burst of rsync traffic from *one* of the houses to the office, but that's going to be as much as it gets - the rest of it is just purely convenience, I suppose.

But yeah, I realise that the problem that the "hidden node" effect creates is that of collision detection being very difficult, as obviously the nodes can't see each other. And yeah, I know that some hardware manufacturers make fancy gear to get around this problem, but it's out of my price range :-) and, given the low speeds we're running at, probably unnecessary.

But for cliff notes, you say that I'll probably be fine for a low-speed network? And, 3 directionals pointing to one omnidirectional is going to be the best option? Probably using a 12dBi omni & 19dBi directionals. As I said, the distances are only 2-3km... thanks again!

Thorn · 09-15-2004

Quote:

Originally Posted by number9

Thorn: Thanks for that. As our upstream bandwidth is limited to 128Kb/s, we certainly won't be doing much traffic. There will probably be a nightly burst of rsync traffic from *one* of the houses to the office, but that's going to be as much as it gets - the rest of it is just purely convenience, I suppose.

But yeah, I realise that the problem that the "hidden node" effect creates is that of collision detection being very difficult, as obviously the nodes can't see each other. And yeah, I know that some hardware manufacturers make fancy gear to get around this problem, but it's out of my price range :-) and, given the low speeds we're running at, probably unnecessary.

But for cliff notes, you say that I'll probably be fine for a low-speed network? And, 3 directionals pointing to one omnidirectional is going to be the best option? Probably using a 12dBi omni & 19dBi directionals. As I said, the distances are only 2-3km... thanks again!

Hidden node is exactly a collision detection problem. For a low-to-mid traffic (not speed) network, most people report that tweaking the CTS/RTS settings overcomes any hidden node issues.

Do the math to figure what antennae are really needed. 12dBi is likely more than enough, and 19dBi sounds like overkill for 3km. YDI has several online calculators for network design.

number9 · 09-15-2004

Quote:

Originally Posted by Thorn

Hidden node is exactly a collision detection problem. For a low-to-mid traffic (not speed) network, most people report that tweaking the CTS/RTS settings overcomes any hidden node issues.

Do the math to figure what antennae are really needed. 12dBi is likely more than enough, and 19dBi sounds like overkill for 3km. YDI has several online calculators for network design.

Well, hopefully the traffic should remain low; as well as the speed.

I mean, noone will be doing actual work from home, which will keep throughput down; and with a slow uplink actual connections to the internet will, hopefully, be light on traffic. During the day I'll probably unplug the AP just to be sure!

I'll see if I can get some lower-power antennae too. I'll check out the YDI site, cheers!

number9 · 09-16-2004

I know that the "Yagi or Parabolic" question has been asked dozens of times, indeed, I've just read 20 threads in the archives on this particular topic! However, I can't actually find a thread that will answer my question, so...

I want to do a point-to-multipoint network, for some low-speed low-traffic data connections (see my other thread asking for advice). I'll use an Omni antenna at the base, but for the clients I can't decide what to use.

I realise that Parabolics usually offer higher gains, but narrower beamwidth (I think?) and thus make for much more difficult pointing. And likewise I realise that Yagis usually offer slightly less gain, but are easier to point. BTW, if I'm wrong about either of this, someone please correct me

The nodes will be around ~3km away from each other, I'm gonna use Senao 100mW bridges on each side. My wifi supplier has stock of 12dB(i) omnis and 19dB(i) parabolics, but I'm not sure if that's overkill? I certainly don't want to be polluting the airwaves, making it harder for other people to use their own connections.

Oh yeah, and I can't even decide between 802.11b and 802.11g

Thorn · 09-16-2004

Quote:

Originally Posted by number9

I know that the "Yagi or Parabolic" question has been asked dozens of times, indeed, I've just read 20 threads in the archives on this particular topic! However, I can't actually find a thread that will answer my question, so...

I want to do a point-to-multipoint network, for some low-speed low-traffic data connections (see my other thread asking for advice). I'll use an Omni antenna at the base, but for the clients I can't decide what to use.

I realise that Parabolics usually offer higher gains, but narrower beamwidth (I think?) and thus make for much more difficult pointing. And likewise I realise that Yagis usually offer slightly less gain, but are easier to point. BTW, if I'm wrong about either of this, someone please correct me

...

That's essentuially correct for your purposes.

(A real "freq geek" would point out there are physics limitations as to why you don't see yagis beyond a certain gain. Dishes are more percise in the design and construction, so are therefore more difficult to make. But because it becomes physically difficult to make a yagi beyond a given size, to obtain higher gains you are forced to dishes.

Huh, having pointed all this out, I guess that makes me real freq geek.

)

Quote:

Originally Posted by number9

... The nodes will be around ~3km away from each other, I'm gonna use Senao 100mW bridges on each side. My wifi supplier has stock of 12dB(i) omnis and 19dB(i) parabolics, but I'm not sure if that's overkill? I certainly don't want to be polluting the airwaves, making it harder for other people to use their own connections. ...

It depends on the gain of the omni and the cable losses, etc. but 12 to 14dBi is probably adequate. However, it is better to jump up to the 19dBi units than to go to RF amplifiers.

Quote:

Originally Posted by number9

... Oh yeah, and I can't even decide between 802.11b and 802.11g

If there is a big price difference, go with 802.11b. On a distance link everything will drop to 5.5 to 1mbs anyway. "G" equipment will work, but you'll never get the "g" speeds.

G8tK33per · 09-16-2004

Quote:

Originally Posted by number9

I know that the "Yagi or Parabolic" question has been asked dozens of times, indeed, I've just read 20 threads in the archives on this particular topic! However, I can't actually find a thread that will answer my question, so...

I want to do a point-to-multipoint network, for some low-speed low-traffic data connections (see my other thread asking for advice). I'll use an Omni antenna at the base, but for the clients I can't decide what to use.

I realise that Parabolics usually offer higher gains, but narrower beamwidth (I think?) and thus make for much more difficult pointing. And likewise I realise that Yagis usually offer slightly less gain, but are easier to point. BTW, if I'm wrong about either of this, someone please correct me

The nodes will be around ~3km away from each other, I'm gonna use Senao 100mW bridges on each side. My wifi supplier has stock of 12dB(i) omnis and 19dB(i) parabolics, but I'm not sure if that's overkill? I certainly don't want to be polluting the airwaves, making it harder for other people to use their own connections.

Oh yeah, and I can't even decide between 802.11b and 802.11g

There is no need to start multiple threads when you are essentially talking about the same subject. In this case, issues concerning point to multi-point setups. Do this outside of the Lounge and your a$$ will belong to me.

Does THIS look familiar?

Thorn · 09-16-2004

Merged.

Novilio · 09-16-2004

One issue that I didn't see brought up earlier in this thread is wind loading. If your antennas are out in the open (rather than tucked under the eaves of a building) you have to take wind loading into account. A yagi is good from that standpoint because it's basically a long rod with some teeth on it. Its wind profile is quite low, and once you point it, it's likely to stay pointed. Large parabolics can have a very high wind profile unless they're skeletal, like the ones you see at FAB. (They look a little like warped barbecue grills.)

I've BSed with guys here who have tried to use old satellite dishes for very high gain wi-fi point-to-point, and wind shoving them off target was a constant issue. It's not for nothing that in satellite service they're typically mounted on a piece of 3" or 4" pipe sunk in concrete.

I've found dishes no harder to align than yagis. Mounting, yes. Alignment is done pretty much the same way, the big challenge being the stability of the mount and the physical difficulty of working with something way up in the air.

One final note: This is a big, complex project, especially for a beginner. Research it thoroughly and do a test connection between your base and one of those outlying nodes before committing a bundle of money to hardware. Especially on the antenna side, like Thorn said, do the math. Calculate your link budget before you buy anything, or you will be sorry.

--73--

--Novilio
(freq geek since 1964.)

number9 · 09-16-2004

Quote:

Originally Posted by Thorn

That's essentuially correct for your purposes.

(A real "freq geek" would point out there are physics limitations as to why you don't see yagis beyond a certain gain. Dishes are more percise in the design and construction, so are therefore more difficult to make. But because it becomes physically difficult to make a yagi beyond a given size, to obtain higher gains you are forced to dishes.

Huh, having pointed all this out, I guess that makes me real freq geek.

)

Ahh.. yes I would say that it would

but yeah, looking at the internal construction of a Yagi it's easy to see how they're more precisely put-together than a parabolic, so I guess it's pretty easy to relate the two & their limitations that way!

Quote:

It depends on the gain of the omni and the cable losses, etc. but 12 to 14dBi is probably adequate. However, it is better to jump up to the 19dBi units than to go to RF amplifiers.

Yeah, probably a 12dBi omni; I'll try to keep cable runs to a few feet if possible. I'm not sure how easy this will be, though? I'm either going to need a weatherproof case, or a 20' cable run. Hmm. It'd be nice if you could scale down power outputs of a bridges, that way you only get what's needed for your link without polluting the airwaves unnecessarily. Oh well...

Quote:

If there is a big price difference, go with 802.11b. On a distance link everything will drop to 5.5 to 1mbs anyway. "G" equipment will work, but you'll never get the "g" speeds.

You're right, I'll end up choosing based on price; as even 1Mb/s will be sufficient for me and I can't see myself pushing much more than 5Mb/s if I get nice connections - I might go G though, given as I'll have other bridges around for spares Just In Case.

Quote:

Originally Posted by G8tK33per

There is no need to start multiple threads when you are essentially talking about the same subject. In this case, issues concerning point to multi-point setups. Do this outside of the Lounge and your a$$ will belong to me.

Apologies for that, because I was asking specific questions about picking between two types of antennas, I thought it'd be best for a new thread. Seeing as I wandered off into the "b vs. g" debate, it's probably worthwhile you kicked my ass

All that said, though, I still don't know to go Yagi or Parabolic! Given the amount of trees and stuff around, and that it's probably going to help me to err on the side of more gain than less gain, I think I'll go Parabolic.

number9 · 09-16-2004

Quote:

Originally Posted by Novilio

One issue that I didn't see brought up earlier in this thread is wind loading. If your antennas are out in the open (rather than tucked under the eaves of a building) you have to take wind loading into account. A yagi is good from that standpoint because it's basically a long rod with some teeth on it. Its wind profile is quite low, and once you point it, it's likely to stay pointed. Large parabolics can have a very high wind profile unless they're skeletal, like the ones you see at FAB. (They look a little like warped barbecue grills.)

Yeah, I know I need to consider wind loading as well. Our redundant ISP connection is actually a parabolic (not sure of the gain, off the top of my head) dish on a high mast on our office roof, and it hasn't moved in the 18 months I've been here. Similarly, we've got a 2.7m VSAT dish that's our primary link to our ISP; and believe me I know how much wind that thing catches!

Quote:

I've BSed with guys here who have tried to use old satellite dishes for very high gain wi-fi point-to-point, and wind shoving them off target was a constant issue. It's not for nothing that in satellite service they're typically mounted on a piece of 3" or 4" pipe sunk in concrete.

Well FWIW, when I had a 2.7-meter dish at the roof our building, the thing was held down with around 360kg (what, 800lb?) of concrete on EACH of it's four feet. So, yeah, construction is important! And yeah, the central mast was at least 4" pipe. I'll take some photos if anyone is interested, wireless doesn't have to mean 802.11x standards!

Quote:

One final note: This is a big, complex project, especially for a beginner. Research it thoroughly and do a test connection between your base and one of those outlying nodes before committing a bundle of money to hardware. Especially on the antenna side, like Thorn said, do the math. Calculate your link budget before you buy anything, or you will be sorry.

Yeah, you're right; it is. At the moment I'm trying to calculate cable runs & associated losses, from there I'll do the sums to see how my signal strength is going to pan out over the link.

Once I'm satisfied that I'm gonna have what it takes for the network to work, I'll source the two antennae and use two AP/Bridges I've got lying around to test. Not that it's my money, but...

Thorn · 09-16-2004

Quote:

Originally Posted by number9

... I'll try to keep cable runs to a few feet if possible. I'm not sure how easy this will be, though? I'm either going to need a weatherproof case, or a 20' cable run. ...

Weatherproof cases aren't difficult to come by (well in the US anyway, no idea for Mali.) Here you can purchase a NEMA4-rated 30cm X 30cm x 20 plastic box for about $25.00.

If you do go that route, which is actually preferable from the standpoint of signal loss, look into PoE (Power over Ethernet.) It is actually rather easy to implement. www.nycwireless.net has a Do-It-Yourself tutorial on constructing them, or you can by premade devices.

LIL_HaKeR · 09-16-2004

Hey Guys My WISP is a local "Neighborhood isp"

I wouldnt mind taking some pics of what we use to connect

09-15-2004	#1 (permalink)
number9 Registered Member Join Date: Jul 2004 Location: Bamako, Mali Posts: 10	Point-to-multipoint basic questions Hi folks, I'm a little new to the whole WiFi world, but for a bit of fun I'm hoping to roll out some gear to do either multiple point-to-point connections or a single point-to-multipoint install. Essentially, I want to connect three LANs (actually, three houses) to our office building LAN (which is one building) in a city. The houses are at most ~3km away from each other, and line-of-sight shouldn't be a problem. I'm wondering, though, how well this is going to work? The reason I ask is that I've heard horror stories about point-to-multipoint and something called, I think, the "hidden node" effect where packets are randomly dropped. Is this true? Bear in mind I have a maximum upstream (i.e. the connection to my ISP) of 128Kb/s, I'm hardly going to be doing high-speed connectivity. I just want to be able to browse web & check my email from home; and given that I live in a developing country this isn't possible with traditional internet access. Any info would be greatly appreciated! Thanks, Chris Bamako, Mali (west Africa) Last edited by number9 : 09-15-2004 at 07:30 AM.

09-16-2004	#6 (permalink)
number9 Registered Member Join Date: Jul 2004 Location: Bamako, Mali Posts: 10	Yagi or Parabolic with P-to-MP? I know that the "Yagi or Parabolic" question has been asked dozens of times, indeed, I've just read 20 threads in the archives on this particular topic! However, I can't actually find a thread that will answer my question, so... I want to do a point-to-multipoint network, for some low-speed low-traffic data connections (see my other thread asking for advice). I'll use an Omni antenna at the base, but for the clients I can't decide what to use. I realise that Parabolics usually offer higher gains, but narrower beamwidth (I think?) and thus make for much more difficult pointing. And likewise I realise that Yagis usually offer slightly less gain, but are easier to point. BTW, if I'm wrong about either of this, someone please correct me The nodes will be around ~3km away from each other, I'm gonna use Senao 100mW bridges on each side. My wifi supplier has stock of 12dB(i) omnis and 19dB(i) parabolics, but I'm not sure if that's overkill? I certainly don't want to be polluting the airwaves, making it harder for other people to use their own connections. Oh yeah, and I can't even decide between 802.11b and 802.11g

09-16-2004	#9 (permalink)
Thorn Did you do the math? Join Date: Apr 2002 Location: Villa Straylight Posts: 10,505	Merged. __________________ Thorn "Read Altas Shrugged. Compare it to today. Repeat as necessary"

09-16-2004	#10 (permalink)
Novilio ScrivenStumbler Join Date: Jul 2002 Location: Colorado Springs, CO Posts: 254	Parabolics and Yagis One issue that I didn't see brought up earlier in this thread is wind loading. If your antennas are out in the open (rather than tucked under the eaves of a building) you have to take wind loading into account. A yagi is good from that standpoint because it's basically a long rod with some teeth on it. Its wind profile is quite low, and once you point it, it's likely to stay pointed. Large parabolics can have a very high wind profile unless they're skeletal, like the ones you see at FAB. (They look a little like warped barbecue grills.) I've BSed with guys here who have tried to use old satellite dishes for very high gain wi-fi point-to-point, and wind shoving them off target was a constant issue. It's not for nothing that in satellite service they're typically mounted on a piece of 3" or 4" pipe sunk in concrete. I've found dishes no harder to align than yagis. Mounting, yes. Alignment is done pretty much the same way, the big challenge being the stability of the mount and the physical difficulty of working with something way up in the air. One final note: This is a big, complex project, especially for a beginner. Research it thoroughly and do a test connection between your base and one of those outlying nodes before committing a bundle of money to hardware. Especially on the antenna side, like Thorn said, do the math. Calculate your link budget before you buy anything, or you will be sorry. --73-- --Novilio (freq geek since 1964.)

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

09-15-2004	#3 (permalink)
number9 Registered Member Join Date: Jul 2004 Location: Bamako, Mali Posts: 10	Thorn: Thanks for that. As our upstream bandwidth is limited to 128Kb/s, we certainly won't be doing much traffic. There will probably be a nightly burst of rsync traffic from one of the houses to the office, but that's going to be as much as it gets - the rest of it is just purely convenience, I suppose. But yeah, I realise that the problem that the "hidden node" effect creates is that of collision detection being very difficult, as obviously the nodes can't see each other. And yeah, I know that some hardware manufacturers make fancy gear to get around this problem, but it's out of my price range :-) and, given the low speeds we're running at, probably unnecessary. But for cliff notes, you say that I'll probably be fine for a low-speed network? And, 3 directionals pointing to one omnidirectional is going to be the best option? Probably using a 12dBi omni & 19dBi directionals. As I said, the distances are only 2-3km... thanks again!

09-16-2004	#14 (permalink)
LIL_HaKeR Registered Member Join Date: Sep 2004 Posts: 14	Hey Guys My WISP is a local "Neighborhood isp" I wouldnt mind taking some pics of what we use to connect