JAWUG

Omni's work. They just work badly for general Jawug networking. The problem is that most people think that putting an omni up at a highsite, and amping it will get you a good connection. Nothing could be further from the truth.

What makes a wireless link work, is the quality of the radio equipment (receive sensitivity), and the least amount of noise possible. The difference between the noise, and the received signal is what determines the quality of your link.

Remember, an amplifier doesn't just amplify the transmitted signal. It also amplifies the received signal, including all the general noise in the spectrum. An omni receives noise from 360degrees around it. Then, when you amplify that noise, and what you are left with is an access point that can be seen for miles, but cannot ever establish a decent connection with a client due to it not being able to differentiate the client from all the noise.

A directional antenna, picks up noise from a much more limited radius (depending on the type) from 18degrees to 60degrees in the case of sectorals.

If omni's were effective, and a good way to build radio networks, ask yourself: Why don't you ever see an omnidirectional on a Cellphone tower? It's because omnidirectional antennas make for bad RF design.

An omni is fine if you're broadcasting something, one-way. Not if you're trying to do highspeed two-way communications.

{joey}
Comments: A directional antenna will also amplify that noise to the same level as an omni would pick up from all directions, i.e. no advantage!

Quote: "Why don't you ever see an omnidirectional on a Cellphone tower? It's because omnidirectional antennas make for bad RF design." Wrong, its because they have the money to buy individual radios for each sector.

They are well-suited for broadcasting, but the argument that high-speed two-way communication is hampered depends on the situation,(i.e. is is point to point or point-to-multi-point?)

pookfuzz+
Ok, rather than remove joey's rather naive observations lets take a real world example. Lets assume I have 2 nodes I need to connect. The distance requires I use a 15dBi+ antenna on each side. I could use 2 omni's, or I could use 2 yagi's. The omni will push out signal in a 360 degree radius, while the yagi will push the signal out over 32 degrees only. The omni produces about 20 times as much rf area coverage as the yagi. Simple enough for everyone to understand I hope. Why 20 times you might ask, well there are two omni's and unlike the yagi there is little common area of coverage between the two antennas.

Oh and just to clear up the cell phone tower story, they use multiple sectors to increase capacity since a single antenna can only service a set number of connections. In areas where there is low customer density you will find they often use omni's.

rodent
heh, this is getting into a rather interesting discussion. Yes, cell providers will use omnis in low density areas. But you will never see them using two, or three omnis, because it doesn't make sense once they start moving to that density.

As to joey's comment about cell providers having the money to buy radios for each sector: I think it proves the point. More radios with directional antennas, provide better performance. WiFi Radios are cheap.

As to a directional picking up the same amount of noise as an omni, that's simply not possible. A directional antenna's design makes it sensitive to RF signals coming from a specific direction. That's why the LNB of e.g. a grid, or dish, is protect from the rear by a disc, or grid. The "faraday" cage at the back, effectively prevents rf signals from entering from unwanted areas, whilst at the same time, reflecting, and focusing received signals from the transmitter onto the LNB.

I cannot possibly see how you can claim a directional antenna picks up the same amount of noise as an omnidirectional antenna.

Let's use an analogy to make it very clear. Imagine you're standing in the middle of a noisy room full of people talking loudly. You're trying to make out the conversation of two specific people on the other end of the room.

If you just stood in the middle, your ears would pickup the noise from everywhere. Now, take a piece of cardboard, fold a funnel, point it at the two people in question, and you'll be able to hear them quite clearly.

The "theoretical" difference between audio waves, and RF waves is not much. They're all waveforms.

Take a look at the radiation pattern differences between an omni: http://www.poynting.co.za/antennas/pdf/omni-a0003_broc_issue_2.pdf and a Grid http://www.poynting.co.za/antennas/pdf/panl_a0004_broc_issue_5.pdf

The same qualifies for the receive signals, as the idea behind a directional antenna is to provide attenuation of interfering signals from areas that you're not interested in.

If you'd like we can do some performance testing, I know someone that has a 30db omni. Gain, is not gain, unless its' from the direction you require. I'll take 24db's of directional gain, over 30db's of omnidirectional gain any time. 802.11 radios hate noise.

Radios are cheap. If you need to cover more than one angle, use two widebeam directionals.

Omni's have their applications. As those little rubber duckies that you get with your stock router. Perhaps Joey can think of another application, for example where you have enough radio density so that distance is not a problem, but that's not the reality for any of the WUG's we're currently building.

Then, let's look at the final logistic: PRICE.

A 15db omni is around R1200. A 13db widebeam directional flatpanel is around R400. A 24db grid is around R700.

So tell me. What are the advantages of Omnis again?

{joey}

In response to pookfuzz+ : If you read my response to the AntennaFAQ you would have seen that I agreed that Omnis are NOT suited for point-to-point links. That is exactly what you described in your "real world application". As for your understanding of RF, can you please explain to me if you included the increase in distance that your Yagi is going to push the signal/noise and "pull" noise from?

If you imagine that the radiation pattern of an omni in the x-y plane (for simplification), the area covered is calculated as:

Area = Pie X Radius^2 (Area of a circle)

Now to extend this to include sectors we say:

Area = Pie X Radius^2 X (Angle/360) (Area of a Sector, where Angle=360 for an omni and Angle=32 according to your yagi)

Now, basic Antenna theory will tell you that if you reduce the angular coverage (i.e.Angle), then your distance will increase (i.e. Radius, R >> r) somewhat proportionally. The equation shows that the change of Radius^2 is inversely proportional to the change of Angle, thus the overall Area should remain the similar.

So instead of "polluting" a large area with a little noise, you are polluting a smaller area with more noise.

In response to rodent:
Multiple radios used with separate sectors provide additional bandwidth and redundancy. They also cost alot more and radios might be cheap to you, but I would argue that it is a relative measure. I don't think that we should marginalise those that can't afford multiple radios from a "community" network by blindly discouraging cheaper alternatives.

In your sound-funnel example, you are simply demonstrating how you can remove point sources of noise from your reception by focusing your receive area. But noise is made up of a noise floor, which is present EVERYWHERE, in all directions and sometimes, in addition, point sources of noise. I garauntee that in your example, the noise floor, present everywhere in the room, is also amplified and thus the ratio between that noise and the sound from the conversation remain the same. I have mentioned before, that if have large point sources of noise coming from directions other than the direction you want to be communicating in, then by all means, use a directional antenna.

In the documentation you refer to from Poynting, the pattern shown for the Omni-directional antenna is in the E-plane, referring to elevation. That shows that the signal is not sent in a spherical 3d pattern, but rather like a doughnut. The diagram does not help nor hinder my or your views on the matter?

As for your cost argument, I can quote you prices of R80 for omnis (ask Ryder) and R1400 for sectors. As far as radios being cheap, R800, that depends on who you're speaking to!

Generally there is a gain vs price relationship and commonly used omnis have lower gain. High gain omnis, like the 15dBi+ you mentioned improve gain by decreasing the elevation angle, so that you end up with a "thinner" doughnut. Incidentally, the lobes also then tend to "bend" downward and as a result and you get even shorter range.

As far as your challenge to do performance testing, I've never stated that an omni will out-perform a directional antenna, especially in a point-to-point setup, because thats clearly not the purpose of an omni. That would be the equivalent of me challenging you to test the performance of a yagi compared to an omnni in communicating with nodes in all directions. But if you want a fair challenge, lets compare your idea of several sectors working off a single radio compared to a single omni on one radio?

I think an important lesson out of this debate is that opinions should be accompanied by a name so that they are recognised as such.

pookfuzz+
joey, if you read what I wrote you would see the yagi pushes the signal the same distance as the omni. Both were 15dBi+.

As for the whole noise issue, may I remind you that ambient noise is subject to the same rule of attenuation. By filtering out the noise sources closest to you, you greatly reduce the amount of noise picked up.

Regardless of what you may wish to believe, omni's when it comes to wifi are not nice. You yourself have admitted they are not good for point to point links and for the meantime these are the type of links that make up the network.

We discourage people from using omni's not due to some odd nazi like intolerance or hatred, we do it because generally when people get omni's they end up unsatisfied with the results and then do stupid things like signal amping. Our opinions are based on real world experiences. You can learn form our mistakes and save yourself the disappointment and waste of money or you can repeat them. Your call really.

{joey}
Quote: "joey, if you read what I wrote you would see the yagi pushes the signal the same distance as the omni. Both were 15dBi++." That statement demonstrates to me a clear lack of understanding of basic RF concepts. There is no direct correlation between gain (15dBi+) and distance. The link equation (look that up) will tell you that the distance depends on both ends of a link. A 15dBi+ omni does not in any way imply that your beam length will be equal to another antenna with equal gain. It can be shown that two 15dBi+ omni's can have different reach distances to another common antenna. The radiation pattern is a 3 dimensional shape, distance is just a measure of one chosen dimension.

If you're not sure about what you're talking about, don't prescribe. Everyone is entitled to an opinion, however, if you want to give advice, advise people to use the right equipment for the situation. Do some research if you're not sure and don't simply rely on opinionated hearsay.

pookfuzz+
You are actually quite funny, from your last statement you want people to believe the the dBi rating of an antenna has nothing to do with the expected operational range? I am sure you could demonstrate minor variances in different antennas but those will be minor and the result of variances in construction. Lets see, output power is pretty fixed, so is rate of attenuation, what we have left is antenna gain and distance from transmitter we can play with, if the antenna gain gets fixed suddenly we only get distance to play with in order to get a usable signal.

As for trying to muddy the water with antennas propagating in 3d, do you realize that this only highlights why omni's are going to cause more rf clutter? With an omni all the energy getting fed in gets pumped out horizontally while with other antenna types a lot of the energy is essentially wasted either from hitting the ground or flying over everyone’s heads. Equal gain antennas push out equal signal densities in their optimal plane of operation, but you don't have to believe me, do the Volts by Meter plots yourself and see.

Personally I am starting to see why rodent has little respect for you guys, I have no idea what you chaps are researching with regards to wifi but I do hope its social in nature because your physical sciences understanding seems very naive. Perhaps I am just not expressing my technical viewpoints well, who knows, I think someone else needs to comment as I get the feeling my attempts are futile or simply not clear enough.

therodent

Ok, I'll change the comment at the top of the page to be: Omni's are bad for general Jawug networking. joey, keep in mind that I also discuss amplifiers in my explanation. This is because this is the most typically deployed kind of omni you will see.

A 30dBi+ omni, with a 500mw amplifier. And those are bad. You keep referring to noise as the noise floor. Sure, there's always base noise. When I refer to noise, I refer to noise from other access points, which invariably affects thruput, and causes CRC errors. And exactly THAT, is what you do NOT want to pick up from 360 degrees around you. When building a WUG you do NOT want to pick up the 10 access points in the street, that are on the same channel. Your definition if "noise" might be yours, and the "technical" description. My description of "noise" is all the junk from the rest of the bluetooth and 802.11 equipment out there.

And real world experience shows that Omnis suck in noisy areas, whilst directional antennae are able to establish very usable connections. You can argue semantics all you want, but I don't see everyone on the WWM switching to omni's soon. If there were enough interest that you could put an access point every 500-800 metres, then yes, omnis might be a nice idea, but mesh networking faces other problems such as latency increasing logarithmically and thruput decreasing logarithmically.

As I indicated earlier, Omni's are not suitable for Jawug networking, as we're building point to point links, aggregating into "highsites". The only reason we're using mesh protocols is because of it's stateful, responsive, and easily deployed nature. It makes building an unmanaged network easy. We're building a network that works because we're interested in building a network that we can use, and performs well.

Whilst the CSIR might be capable of putting a room full of WRT's together and make them mesh using little omni's, it does not match the real world. The real world situation is that none of the Jawug members are capable of networking efficiently with omnidirectional antennae. It simply doesn't work over the distances we're taling about.

Many of the core members are RF and 802.11 network builders for a living. They have experience on the field. The opinions are not based on hearsay. It's based on real world experience.

{joey}
In reply to pookfuzz+, dude, lets just agree to disagree then. You have your opinion and I have mine. If you don't respect that people can have other opinions than yours, than I don't respect you much either Im afraid and don't see the point of debating further.

In reply to rodent
I'm glad you've taken the time to clarify what you mean and I agree with alot of it, especially with regards to amplification. Furthermore, I don't want to see everyone in WWM using omni's, because that would be senseless, however, if someone feels that an omni might best suit their situation (like a dense part of the network), they should feel free to try and not feel that the JAWUG "elders" will be offended.

Quote: "If there were enough interest that you could put an access point every 500-800 metres, then yes, omnis might be a nice idea" Thats all I wanted at the end of the day. Rather than take a blanket view that omnis are bad, admit that they have a place(however minute that may be in the current JAWUG reality), so that if and hopefully when the situation arises( we want the network to grow dont we?), people can make their own decision.

Quote: "Whilst the CSIR might be capable of putting a room full of WRT's together and make them mesh using little omni's, it does not match the real world.."
How can you express an opinion on what we do without actually coming to see for yourself? You've carelessly ranted on about the CSIR on several occassions but you've never actually taken the effort to come and see for yourself. You're ignoring my invitation to set things straight. Are you scared that you might have to admit that you are wrong about the CSIR?

rodent:
I received your invitation yesterday, after being on holiday for a week after the conference. Mail piles up, and it's been slow wading through it all. I think I'm perfectly allowed to express an opinion of the CSIR with the information that's available to me, a member of the public.

I don't carelessly rant about the CSIR. I look at the CSIR from a taxpayers perspective, and from the perspective of someone that attended the first opensource conference in South Africa, and hearing about "meraka". "Common grazing ground" indeed. It's more like the CSIR's private little greenfields. I look at the CSIR from the perspective of the promises that were made back then by the management of the CSIR at said conference, and the results that have been produced in two years.

I look at the CSIR as someone that's saying: "Where's the transparency?" Your invitation is the first real world transparent thing that the CSIR coin project has done in terms of telling the world what you are doing, aside from a smallish press release about cantennas in Peebles valley late last year.

I look at the CSIR COIN project and I have to ask - why aren't you vocalizing the fact that the Peebles valley network you built is illegal? Why aren't you lobbying for the freeing of WiFi ?

The CSIR wirelessafrica site was launched in July this year only, yet the COIN project has been going for nearly two years now? I think the skepticisim is warranted when one considers that all that the project has done so far in the public's eye is build cantennas and made use of OpenWrt and OLSR. Where is the public version of GISRAP? Why does the CSIR charge for access to cartographic data? I rant about all these things because the CSIR, in the public eye, hoards information, and then charges people for it.

I built a cantenna too, but don't claim to have invented it, and didn't create a press release about it. Jawug started in 2003, with zero funding, and nobody working on it fulltime. The "elders" as you refer to them, are just people who are enthusiasts and had the experience and will to have brought the network to where it is, and have donate an enormous amount of time, and money towards the goals of community networking. As you can see, we're open to debate. We're transparent. The CSIR appears unapproachable and unwilling to share. Perhaps together we can change that perception.

You come here and critique us on our website (which is editable by anyone bothering to register), yet the COIN site doesn't allow open editing or contribution by anyone else, so that we may critique you in your own back yard, or perhaps (gasp!) contribute towards the COIN project, or - make use of some of the CSIR's precious bandwidth to store and save files!

I really look forward to seeing what the CSIR COIN project is doing that WiFi enthusiasts haven't already done twice, and I will definately make use of your invitation.

{joey}
Great, thats excellant. I look forward to you expressing your critism as constructively as you just have, to my team, so that we can answer your questions, correct some of the misconceptions,as well as work on our shortcomings.

Some of your misconceptions stem from the fact that The Meraka Institute, launched only this year, is not the same Meraka you heard about at the Open-source conference. Meraka was a small centre within the CSIR trying to promote its Open Source initiative. The Meraka Institute on the other hand is the result of a National strategic initiative to create an ICT research mass in Africa. The staff of the ICT research centre of the CSIR, ISTC were combined with the Meraka you know, to form the core of the Meraka Institute.

As far as my critique of your website, I thought that this is a community network? That to me would imply that anyone who is part of this community is allowed to have an opinion, regardless of whether it goes against the webmasters opinion? I have always made comments as a member of JAWUG, because I feel that I am part of this community and have made some contribution to it. It was only in response to your and pookfuzz's critism of my work that I reacted as an employee of the CSIR.

I truly believe that you and the core JAWUG have done exceptional work in creating this community. I would not have realised this had I not taken the initiative to join you and see exactly what it is you're doing before making comments. By the way, that is also what led me to believe that both JAWUG and the Meraka Institute can truly benefit from some dialogue.

See u at the CSIR

int03h
SHIELDS UP - DISCLAIMERS ON: I think the discussion above is a symptom not the cause. I think alot of people (put yourself into whatever category you see fit) believe that a community wifi network is not subject to rules or regulation. I think it is more likely to go "titsup" because there are no rules. If its a free for all .. then there is no sanity to prevail and we all end up buying omnis with the biggest baddest amps so that we can steal more of the available bandwidth from our next door neigbour. Likewise the guy down the street is going to have to buy a bigger badder setup to achieve his goals and talk over me and in the end we all end up with no reliable comms.
IMHO the WHOLE of the above discussion are guidelines for creating a stable network with the technology available to adress the particular challenges ( economic, distance, time, technological, REGULATORY, competitive etc). I think it is a majority rules and the history will shape the future type of thing. When new technology becomes available (CMCA type network-i.e. better time slicing, medium sharing, noise prevention rather than noise handling) and the REGULATORY environment becomes more open, this discussion will certanily be different in terms of the appropriate choice of antennae. Personally I don't see this as being that far away, 2 years tops... I see this as a learning curve for the future.. I personally don't believe 802.11x technology is ready for mass deployment in SA. But it is great to play with now! At least it puts the community in control of the destiny of OUR network rather than allowing a space for others to commercialise and ultimately regulate it.
And likewise .. see you at the CSIR.