Connecting 2 Wifi Networks Via Cantenna

On 30 Apr 2007 11:05:10 -0700, snipped-for-privacy@att.net wrote in :

You need:

1. Wireless Ethernet client bridge with high-gain directional antenna to connect to your neighbor; connected by Ethernet cable to:
2. Wireless access point (or wireless router configured as an access point as described in the How To wiki below) on a different primary channel (1, 6, 11) with a different and unique SSID.

What John said.

First question is if your neighbor's AP/router can have an external directional antenna. Is he using it for wireless there or just as a router? If he's not using wireless in his house, then connect a cantenna or a panel antenna or a yagi via a short as possible low-loss cable to that router. Point it towards you.

Next, whether you can get that directional antenna or have to stick with the omni that came with his, you definitely want a directional antenna on your end to connect with him using a wireless client bridge

The Buffalo WHR-HP-G54 would be a good consumer level router that can also function as a client. BestBuy or NewEgg will have them. They are popular. The HP is preferred, for it's "high-power".

Again, be carefull about cable loss. It's critical. If possible, put the client bridge close to or directly connected to the antenna. I generally avoid antennas with their own cables unless they are short and LMR200 or better. This calculator helps you decide the cable part:

1) your pc and be done but not wireless in your house. 2) another router set up as an AP

Another option if wired is ok is to buy a USB adapter with external antenna jack and put that same directional antenna on it. In this case, it will only serve for one pc.

To determine what gain you need for your antenna(s), read the wireless wiki in John's signature and check out the link calculations for standard operating margin, SOM. You should aim for 20db minimum of signal to noise ratio. u

Cantennas may be great, but you can find hi gain antennas online for just a little more.

Cheers, Steve

An aside: Once you get hooked up, make sure you understand the FAP policy with the Satellite connection. It is restrictive and can be punishing if you are not careful with downloading.

The great thing is that with two of you, you can upgrade to a higher service level without too much pain.

Steve

snipped-for-privacy@att.net hath wroth:

See comments by John and Seaweed for what you need. However, be prepared for a problem. My guess is that the satellite is either HughesNet (Direcway) or Wild Blue service. Both have severe limitations on use called FAP (fair access policy). I know of several small groups that are sharing a satellite connection. It takes considerable communal effort and some yelling to make it usable. There are also some service limitations. VoIP is half duplex with about a 2 second delay. It works but takes practice learning to say "over" at the end of each "transmission". Uploads are incredibly slow. VPN's are a lost cause. I suggest you try the neighbors satellite system before you spend much time and money.

Also, there are often better ways to do this than wireless. If you can run cable across the 1/4 mile between houses, methinks it's better to run fiber, coax, and twisted pair, than to rely on wireless. Wireless is fine, but is subject to interference problems. If you have a bunch of nearby Wi-Fi networks, this may be a problem. Details of various methods on request.

Thanks for all the tips folks. I may actually get my own dish, but I have several other projects brewing that require the same type of setup including a robo mower to cut the 6 acres that I currently mow. I'd like to be able to use winsock for control and monitoring as well as a DGPS setup. In addition I have city rental properties that are subject to frequent vandalism. If I can get a sat or cable connection at one site and beam wifi around to the other sites using directional antennas I can use it for monitored alarm systems and surveilance cameras. This one is probably going to be more challenging than it is worth, due to obstructed line of site issues. But maybe not completely as I have a few properties that are near each other. Just the same I'm considering alternatives such as trac phones or cell family plans. These systems will only have to send a heartbeat (I'm still here) for monitoring and then data in the infrequent (hopefully) alarm event.

I'll be reading the FAQ's and may have more questions.

Thanks again!!

Good advice and good products. I use them on my boat with good results, sometimes over 3 miles.

The trick here, if it's Hughes is to combine your resources to get a business-level plan. This actually makes it better to share a dish. For example, 1 person on the Hughes Home plan gets, supposedly

200 MB a...um, let's say day, at "up to" 700 Kbps and 50 kbps recovery for $70 a month. Upgrading to the Small-Business plan, for $100 a month you can have more than double all those numbers for $50 each.

And it's better than that, because to the degree that you don't overlap in usage, each of you can enjoy the full bandwidth to yourselves at times. Especially if you coordinate/cooperate. We have

8 people (obviously not power users at all) sharing that $100 account. We may add two more and move to the $200 account, but that's still cheap !

up and share a bigger account/fatter pipe. It's like the difference between running one 2" water pipe to your neighborhood to service 20 houses vs running 20 1/2" pipes. Which is more efficient? Which gives better pressure/flow?

Still won't help for VoIP or VPN no matter what account you have.

Note: FAP for Hughes net is extremely complex and had just changed this month. Also, they are not disclosing any real numbers. Look to the sat forum for more info.

Still Hughes is probably the best of the consumer level systems.

Steve

I don't know why I keep getting those arrows > > inserted in the middle of my messages. Makes it hard to sort out who is saying what.

Steve

seaweedsteve hath wroth:

If you include wall drag (Bernoulli's equation), then the plumbing system with the smallest wall area gives less pressure loss:

This assumes incompressible fluid flow, which isn't a very good analogy for typical web data traffic.

I sorta help with 3 different systems that share a satellite link. All are HughesNet (DirecWay). The planned Wild Blue system fell through. The problem is that all 3 of these have DSL or cable available within wire or wireless range. However, they all have also successfully precipitated a war with the neighbors making a wired or wireless link somewhat problematic. I've been elected official diplomat and am trying to negotiate a peacful compromise that will allow them to obtain a terrestial alternative. However, no sooner do I obtain some sort of verbal agreement, someone always manages to restart the battle. I'm scheduled for yet another non-billable endless debate this weekend. Sigh.

I've suggested getting wider bandwidth connections. One of these has just done that. I haven't seen the results. The others are still waiting for me to "somehow" provide them with a terrestial solution and therefore don't want to invest in a 1 year HughesNet commitment.

Yep. Physics can only be overcome by politics.

Ouch. I hadn't noticed. Kinda reminds me of trying to come up with the ultimate fair equitable and manageable ISP billing system. Little wonder the industry has almost universally gone to flat rate. It's really difficult to be fair and equitable.

OK. You got me there. You're not supposed to take it too literally ! I was trying for an analogy to express the fact that by ganging up users you get a combining effect. For example, if one user needs a "700 kbps" connection for a "high-speed" feel, 6 users don't need 4200 kbps, but will have a good experience with 1500. Of course, I'm talking about the touted "up-to" speeds. The delivered speeds are often 1/2-3/4 that.

Um, political problems. Another can of worms. As we all know, satellite is only for those who can't get broadband. Or even medium band. But it's better than dial-up still. Unless/until you hit FAP, that is!

I think that they will like it better. And with the new FAP, it may make the critical difference on being able to watch youtube, whatever. Of course, if it's a bad installation or small-dish, then it may not help the speed. My experience is that the real difference is going to the 2-watt transmitter and big dish acounts. I think you get priority as a biz class user.

Huh?

I'm wondering when they will start selling bandwidth like electricity. Especially on Satellites where it's so expensive for the providers to acquire. Put a meter on the sucker and charge by the gigabyte. Then they will need no FAP. Problem is, I think that most users consume very little bandwidth and are subsidizing people like me. And the bigger bandwidth users would drop way back if they had to pay what it really costs on Sat.

I suppose that just like electricity and water, you simply charge a minimum fee for the connection and then charge usage for anything over that.

Steve

On 4 May 2007 07:14:14 -0700, seaweedsteve wrote in :

The rule of thumb used to be 10:1 (overselling bandwidth by 10x), but all it takes is one bandwidth hog (e.g., file sharer) to invalidate that.

They've tried that, but the market hasn't accepted it, particularly when the competition (DSL, cable) are selling "unlimited". So they lie instead.

seaweedsteve hath wroth:

Sorry. I was just trying to secure my pedantic hair-splitting nit-picker certificate. Much as I really like analogies, I also enjoy shooting them full of holes. In the end, all analogies fail.

I wish it were that simple. I have a few customers with multiple DSL connections. They use various load balancing routers with mixed results:

Some things work really well. Others (large single-threaded uploads) aren't worth the effort. Incidentally, a plumbing analogy of one of these load balancing routers fails almost immediately.

Well, it might be nice if they went for the business class service, but the new equipment, higher bill, and large number of shared users, makes it a dubious proposition. They currently have a small 0.74m dish, and low power xmitter. At least they have a decent DW-7000S(???) router. I'm pushing for a terrestrial solution, but have to make peace with the neighbors first. Incidentally, my visitation is postponed thanks to me catching a cold or flu.

I have to deal with the FCC and a few politicians all too frequently. Politicians are the only ones that will consistently ignore technical advice, industry expertise, and basic physics. They will barge ahead on some pork barrel project knowing full well that it won't work.

Metered service has been tried many times and always failed as long as some other provider is offering flat rate. As your abusive bandwidth consumption testifies, there are benefits to having flat rate service. There are also benefits for the ISP, where there's no need to explain, or calculate bandwidth use or consumption. I still remember the bad old daze of Compuserve, where I paid by the minute. Surfing with a stopwatch was common.

Anyway, you currently have the best compromise. Tiered service with abuse management. The bandwidth hogs get to pay more, but not so fine grained that it creates billing debates. When the frequency bands fill up, and the ISP's are forced to provide "priority" service to those that are willing to pay, then we'll have tiered service.

I don't want to go into detail, but I've thrown together several ultimate fair and equitable billing systems that failed miserably. It's far more complicated than you can ever imagine.

John Navas hath wroth:

Chuckle. I think you'll find that the satellite uplink path is far more oversubscribed than just 10:1. If you want, I'll dig out the number of transponders, their aggregate bandwidth, and divide it into the number of satellite users (all of which have essentially full time connections). Actually, I'm not familiar enough with the modulation methods to make an accurate guess, but a ballpark guess might be possible. DW has 260,000 subs

scattered over who knows how many birds and transponders. Upload bandwidth is 50Kbits/sec on a 6MHz wide transponder.

The number of transponders is a bit messy. See:

I went to the PDF list of transponders and simply added up all the bandwidth available on all "broadcast (downlink)" transponders. I get

1,784Gbits/sec download available on 11 birds.

If every customer started downloading simultaneously, each would get: 1.8Gbits/sec / 260,000 customers = 7Kbits/sec which isn't going to work very well.

If every customer were guaranteed 500Kbits/sec, then the system could theoretically handle: 1.8Gbits/sec / 0.5Mbits/sec = 3600 customers which is also not very good.

The oversubscription rate would therefore be: 260,000 customers / 3600 customers = 72:1 oversubscription

I'm too lazy to do the upload numbers, but I suspect they're in the same ballpark (or worse).

I forget exactly, but some of the sat experts have thrown around similiar numbers. 5Kbits/sec per user is what stuck in my head. Thankfully, most of the connections are idle most of the time.

Yep. Or more.

As they improve the ways that they can cram more data into the same bandwidth they are certainly increasing how many they can get on each transponder, although perhaps the effective contention ratio is maintained or lowered. That's why they encourage subscribers to go to the new HN7000S (previous was the DW7000). It doesn't actually improve the subsciber's experience, but when pointed at the right satellite running the new...protocol(?), it allows more to be packed in. I forget the terms, something about slicing each rotation of something into four parts instead of two or one. This is good for the user simply because if Hughes can cram more on each transponder then they can turn a decent profit and won't have to clamp down on bandwidth instead.

One can imagine that it was easy five years ago to offer x amount of thruput, knowing few would use it. But with P2P, youtube and itunes type uses on the increase, they have to clamp down just to maintain the same numbers of accounts.

I've heard that upload is worse. Don't understand why. Probably just because it's bandwidth that doesn't get noticed, therefore doesn't pay off.

Steve

Of course, but to the degree that they cover the main points and help us understand, they are valuable. Sort of like the analogies they use to explain electricity like water. Doesn't go far, but it gets us beginners over the first hurdle of understanding the basic behaviour.

Wow, that is a much more complex kind of sharing. I'm just talking about putting 8-10 people on a home router off of a hughesnet modem. No balancing needed and it's working ! Most of the internet shops around here do exactly the same thing. Not complicated.

.....

Ah, now I understand. I have a family member who is an AI expert. He goes/went to Washington periodically to sit down with other experts and generals and politicians and assess computerized systems for managing big government things, like say weapons or I don't know what, he doesn't say. All he says is that when they ask if this new system will work, if it is safe, and everyone says, "oh, yes" and he says "well, these systems have been known to fail, and maybe it's not a good idea... and ... " Well, they don't want to hear it. Better to find a different scientist. Maybe he's been replaced by now...

As I think about it, I can already see where it gets problematic. Bandwidth is much more burst-able and also more subject to "prime-time effects. Just selling by quantity does not help with the problem of everybody wanting to use it at the same time.

And working in a billing system that dynamically charges for the contention ratio at the moment would be difficult to defend to the customer. They may work it out so that the more you use, the slower it comes, but that's starting to sound a lot like what they are doing already. I think that you are right, it's already perhaps the right approach, though it needs tweaking. That's for sure.

Steve

seaweedsteve hath wroth:

Your 5KBits/sec per sub would probably be more accurate. The list I used included birds that are not available everywhere. Some transponders will be more crowded than others. It also assumes no compression, which might increase the numbers slightly. However, as packet size gets smaller, compression is often self defeating.

As for most connections being idle, that's a good assumption for normal use. However, HughesNet is advertising heavily in the emergency services sector. I expect to see rather heavy priority traffic from those systems during disasters and drills.

My guess(tm) is that they've done just about all that's possible with protocol efficiency, compression, and adaptive bandwidth control. Currently, everyone seems to be working on predictive encoding (similar to what's done on hard disks) and router in the sky switching. However, that won't get you less latency or better thruput. It will just cram more customers into the same transponder bandwidth.

Incidentally, wholesale satellite transponder bandwidth sells for about $4,000 to $7,000 per MHz per month:

in case you want to go into competition with HughesNet.

Commercial satellite bandwidth is often sold on the basis of contention ratio (channel loading). My 72:1 guess is called the "contention ratio".

It also applies to any shared medium such as cable modems and DSL.

I'll plead partial ignorance, but will do some reading on the subject. Kinda sounds like what the WISP vendors found out 10 years ago.

CSMA/CA (carrier sense multiple access / collision avoidance) system fall apart at high channel loading. That's where rotary or dynamic polling works much better. My guess(tm) is that Hughesnet is going to polling. I wouldn't be suprised if they implimented some kind of QoS system in the router to delay packets that are not time critical. Dunno, and just guessing.

Well, I was going to grind the numbers, but don't have enough info. The chart has the transponder ID's, but does not indicate the upload bandwidth. I can probably guess, but it's easier to ask. Later, maybe.

On Sun, 06 May 2007 15:29:43 -0700, Jeff Liebermann wrote in :

That's not what I meant. My oversubscription rate applied only to the "last mile" as in the case of a multiplexor. As you go upstream on service tiers (e.g., neighborhood service, backhaul), the oversubscription rate increases, and 100:1 isn't uncommon (or unworkable). Satellite is a special case because of the lack of service tiers.

Those are different animals. Cable is a shared medium, but DSL isn't, except in the sense that _most_ fat Internet pipes are shared; e.g., dedicated business T-1 aggregated onto shared Internet backhaul.

Polling would almost certainly be a killer with satellite latency. What makes much more sense is time slot pre-assignment.

That could be, probably based on FAP and traffic level.

Guessing in my case because I haven't looked hard at what is being some in a number of years.

On Fri, 04 May 2007 14:31:35 -0700, Jeff Liebermann wrote in :

Maybe, but I doubt it. Capacity appears to be increasing more rapidly than demand.

Did you see my calculations of capacity and user count for HughesNet?

HughesNet has announced plans to increase capacity by adding birds and juggling protocols,

However at 72:1 oversubscription or worse, it's going to take quite a bit of added capacity to get the oversubscription down to the terrestrial standard of 10:1. From my position on the ground, subscriber count seems to be increasing, without the any corresponding increase in available capacity (bandwidth).

However, I goofed. I was about a year behind in my numbers. HughesNet now has 325,000 subscribers.

Using the previous assumption of 500Kbits/sec per subscriber, I would guess: 325,000 / 3600 = 90:1 oversubscription

Now, where is this increasing satellite capacity you've seen rapidly increasing or are you referring to terrestrial fiber capacity?