Wireless Access point

I have a Belkin F5D7230-4 router and the Belkin F5D7010 NIC. This is a 54g setup. I am looking to extend the range of the network by adding an access point. in looking to eBay for a Belkin access point I alos see other brnads of "54g" routers and access points. My question is, will these work with the NIC card I have? Will I get 54g speeds if I get say a LinkSys or D-Link instead of a Belkin with the Belkin NIC?

Thanks for any info...

Mike...

Reply to
Mike Benz
Loading thread data ...

Yes they are interchangable, but why are you concerned about the NIC. You will want to connect the access point to the router not a NIC.

Reply to
Robert Jacobs

other brnads

The f5d7230-4 has WDS capabilities, if you get another belkin like a AP f5d7130 that also has WDS, you could set the AP up as a repeater an not have to have any wires connecting it. Repeater mode will give you about half of your current throughput. If you buy a linksys or dlink AP then you will most likely have to have a ethernet cable attatching it to your f5d7230-4.

Reply to
Airhead

Mike,

You are aware that the 7230 itself acts as a router, access point, and also can even have "bridging" enabled to give you two access points on one network, correct? That's what you'll need to do if you really want to "extend the range of the network".

I have five 7230's here that I've ordered within the past week. I'm chaining them all linearly in the "Wireless Bridge" table in the setup where you enter the MAC addresses for any bridge mode Belkin devices that you want to setup. Configuration in general for this is a beatch.

IMPORTANT NOTE: The Belkin 54g routers will only attach to other Belkin devices for this "bridge mode":

formatting link

Reply to
Bill Evans

In the five Belkins I have chained together, I see a 5% drop off as you go through an additional device.

Reply to
Bill Evans

That's not the way it's suppose to happen with WDS. Each half duplex store-n-forward hop is suppose to cut your bandwidth in half. What might be happening (my guess) is that you're possibly measuring your download speed from the internet. The 802.11g wireless connection is MUCH faster than the typical broadband. Typical is 22Mbits/sec thruput. If you're getting perhaps 1.5Mbits/sec from a DSL or

3Mbits/sec cable connection, you'll never see a drop in bandwidth via a WDS repeater chain. However, if you use another computer at the end of the chain, and use either a benchmarking program, download meter, or simple file transfer to measure performance, you'll probably see the drop.

However, even with 5 hops, you should see a 2^5 or 32 times drop in bandwidth with WDS, mesh, or repeaters. See:

formatting link
#2 and #4 for how it should work.

I've only seen the effects of one WDS hop and it was half. I've never tried to do more than that except in a mesh network with a maximum of

4 hops. I've also done it with the old Metricom/Ricochet system and noted the bandwidth drop as the data went through multiple poletop repeaters (on the old single frequency poletops, not the newer dual frequency poletops).

So, what did you do to get only 5% total? Does your Belkin do full duplex on different channels in order to eliminate the half duplex problem?

Reply to
Jeff Liebermann

throughput.

duplex

Heres an interesting article about some wds comparison test. It shows the first hop at about half or less, then each hop after is not quite as detrimental.

formatting link

Reply to
Airhead

Yes it does, I dont quite understand the relation between the overview and the actual test quoted further in the article.

Reply to
Airhead

That looks like more than 50% drop on each hop.

"Using the network topology described in the Overview section and Intersil's Benchmark Pro 4.7 software run from the LAN PC to each Station, the following sustained TCP throughput performance should be observed, without any other traffic on the network, in a 2.4 GHz interference-free environment: 4.5 Mbps for 1 hop 1.8 Mbps for 2 hops 0.7 Mbps for 3 hops "

Reply to
dold

Why? Because it was full duplex before it was stored and shortly later forwarded?

All true, all of that things I knew, but it wasn't clicking in my head. It was more of a grinding sound, but I digress....

What I'm doing or trying to do is bridge five Belkin 54g's across a building. I could have run a cat 5 cable across the thing, but ethernet cable is 28 cents a foot, and 54g's are $32 at Buy.com. I think that attacking this problem this way is also much more useful in this situation than a naked cat 5 cable would be.

< This network setup is explained in more depth in another thread I started on 2/2, titled "Range on a pair of Belkin 54G routers in wireless bridge mode?". In particular the last, followup post explains the layout of the site. >

These routers are going to be going into a local country club, where they'll basically be spanning the dining hall and the bar. Instead of taking them out of the boxes on site and trying to figure out how to set them up, I've been doing all that here at the house.

So far this is what I've got: I have three of the 54g's set up in bridge mode, with one controlling the whole network and the other two acting as slaves to the first.

I've attached an AutoCAD drawing I just made to describe it to you. I also sent it in WMF format.

I have no idea.

I tested nodes two and three last night, and #2 was around 680 and the other

650. If you'll look at that pic I attached, node three is running FASTER than node two for some reason.
Reply to
Bill Evans

Nope. Let me try to explain... but first a diversion.

There are a few point to point wireless bridges that claimed to be "full duplex". As far as the ethernet port was concerned, the are full duplex because they can support most of the NWAY negotiated full and half duplex 10/100 protocols. However, they are not full duplex as far as the radios are concerned. The way these work is that the radio is half-duplex (transmits or receives, but not at the same time) running at TWICE the speed of the interface. So, if the wireless bridge claims that it can do 3Mbits/sec full duplex, then the radios are running at 6Mbits/sec thruput, with a very fast turn around time to make it appear to be full duplex.

There are also a few radios that are really full duplex. Proxim Lynx and similar wireless bridges are a good example. They use half the

2.4 or 5.6Ghz band in one direction, and the other half in the other direction. These can and do transmit and receive at the same time (full duplex) because each direction is on different frequencies.

The common 802.11a/b/g is half duplex, it can only transmit or receive, but not at the same time. There are mesh systems that use two radios at each node. These are full duplex because each radio is on a different frequency (or channel) allowing the system to send and receive at the same time.

End of diversion and enter WDS (wireless distribution something). WDS is sort of a combination of bridging, access point protocols, mesh networking, and store and forward repeating. There's lots of overlap and I'll try to be consistant in my abuse of the terms.

WDS turns an access point into a combination access point and wireless bridge in that it allows access points to talk to each other. This is normally NOT the case without WDS. Light reading:

formatting link
reading:
formatting link
Let's simplify your specific WDS arrangement so I can explain how the bandwidth reduction is suppose to work. To do that, I need to make a few assumptions, which may not exactly be true in your installation:

  1. All 5 of the WDS access points can sorta hear each other but some are more noisy than others.
  2. Each WDS access point has a computah plugged into the LAN port to avoid the complications of an additional wireless hop to a wireless client.
  3. The topology is arranged so that each AP is configured to accept packets only to/from its immediate neighbor. This is NOT a mesh where every AP can accept packets from every other AP.

AP#1 gets the full bandwidth from a DSL or cable modem connection. A computer plugged into the LAN port will also get full speed.

AP#2 will full bandwidth because the wireless is much faster than the DSL, and each packet need be sent only once from AP#1 to AP#2.

AP#3 has a problem. It gets its packets from AP#2 only. It can hear AP#1 but doesn't do anything with the packets. (Note: this is not true for a mesh network where every node can usually accept packets from any other node). AP#1 sends a packet that's picked up by AP#2. When done receiving, AP#2 retransmits the same identical packet to AP#3. In other words, you have two packets in the air, occupying the time it would normally take to deliver two packets, but only delivering one packet to AP#3. That cuts your bandwidth in half.

Now, AP#4 gets all its packets from AP#3 and has to wait until AP#1 and AP#2 are done passing the packet to AP#3. When that finally happens, AP#3 can send the same packet to AP#4. Meanwhile AP#1 and AP#2 just sit there because they cannot transmit at the same time when AP#3 passes the packet to AP#4. This cuts the bandwidth in half again because AP#1 and AP#2 are effectively idle while AP#3 and AP#4 do their thing.

AP#5 is similar to AP#4. It gets its packets from AP#4 only. When it gets its packet from AP#4, AP#1, AP#2, and AP#3 are just sitting there doing nothing again causing the bandwidth to get cut in half.

So, for 5 hops, we get 2^(5=1) or 1/16th the bandwidth at the end.

Now, lets look at my assumptions.

  1. All 5 WDS access points probably cannot all hear each other. If the AP#1 to AP#2 end cannot hear the AP#4 to AP#5 end, then both of these can transmit at the same time without causing a collision. Therefore, the bandwidth reduction will not be exactly halved for these.
  2. Plugging computahs into the LAN ports eliminates the added delay of using a wireless client. Since the client has exactly the same half duplex problem as the access points, it adds an additional 1/2 reduction in bandwidth. Therefore, a two access point WDS system, with a wireless client connected to AP#1, will result in a 50% bandwidth reduction. Same with any AP along the chain. If connected to AP#5 at the end, the bandwidth reduction is 1/32.
  3. Arranging the topology in a series connected string is an artificial handicap. If the access points can all hear each other, then there's no reason why the packets have to go the long way around. A packet can simply go directly from AP#1 to AP#5 and incurr no bandwidth reduction instead of a 1/16 reduction. That's the way a mesh network operates. There are other problems with a mesh, but I don't wanna get into that now.

Just a note on cable prices. I pay 7 cents/ft for CAT5E in 1000ft rolls and I buy the best (Belden). Other brands can be had for 5 cents/ft. I have paid 20 cents per ft but that was for shielded and gel filled outdoor CAT5. CAT6 is quite a bit more expensive, but you're not running gigabit ethernet beteen boxes, so it's not needed.

I beg to differ. The added delays, bandwidth reduction, airtime use, and configuration complexity offer more complications than solutions. Think of it this way. Your allowable airtime in a given airspace is limited. Only one radio can be transmitting at the same time in the same airspace on the same channel. Would you rather have the airtime used by a wireless client moving data from a local accesss point, or would you prefer having the airtime consumed by WDS bridges sending the exact same data to each other? I would go for the wiring if possible.

Incidentally, I've run 10baseT-HDX over 25 pair telco cables for about

100ft without much difficulty. Don't tell anyone I suggested it.

Argh. I'm out of time... I'll dig for the layout details later.

An excellent idea. I try to simulate my installations before I arrive.

OK, you've built a star topology, not a linear chain like I previously described. You'll get should get no bandwidth reduction to any of the three AP's for directly connected LAN computers. You'll get a 50% reduction for wireless computers connected to any of the three AP's. Traffic from wireless laptop to wireless laptop will depend on how you have the star configured. If the traffic has to go through three (two slaves and one master) to go between laptops, you'll have 1/8 the bandwidth. If the slaves can talk to each other, then it will be 1/4.

Ummm... I have an ACAD viewer in the office. I'll look at it later. A JPG or GIF would have been easier.

I do. You answered the question with the topology. If you measured the bandwidth at any of the three slave AP's with a wired LAN connected computah, you'll see no bandwidth reduction. Try it between LAN connected computahs and you should see a substantial drop in bandwidth.

Bits or bytes?

Interference from the other radios? Try turning off the unused radios and AP's. With a star type topology, it should be identical at all the slaves. How many radios total (including laptops and clients) do you have running in a small airspace?

Reply to
Jeff Liebermann

These Belkin 54g units (they call both the 7230's and the 7130's "54g's") are WDS, and apparently WDS stands for Wireless Distribution System.

This building would be a real beatch to get into the attic or wall space to run cable, and the other building that I'm trying to link to has ONE computer that'll be using the internet.

Even knowing how badly it will hurt the bandwidth on that one machine, I'm planning on going ahead with the poorly designed (on my part) series of Belkins in bridge mode.

I'm confused, all three AP's that I have set up here at my house have wired LAN connections to computers, is that paragraph above what you meant to say?

Bits.

I just reran the tests on all three machines connected to three different Belkin 7230-4 routers, and they got these results in order 1652.5, 314.8, and 688.8.

All numbers are kbps, and my ISP is 3 mbps Charter Cable.

None. Really.

Reply to
Bill Evans

I can never remember if it's "service" or "system". The "service" version of WDS were the 2.1GHz frequencies assigned to Metricom for poletop to poletop backhaul.

Well, ok. That's a good reason to use WDS. However, I've often used existing telephone and coax cable wiring as a backhaul with good success. For example, these can be used at 10baseT-HDX speeds (not

100baseTX) without much difficulty. The 100meter limit for 10baseT also doesn't really apply when there are only two endpoints, no taps, and switched ports at both ends. I've gone 950ft this way over RG-6/u 75ohm coax this way. If the club house is wired for CATV, you can also "borrow" the coax runs using "adapters":
formatting link
formatting link
I suggest you explore some alternatives which might be useful.

No problem. I just wanted to be sure you understood how the bandwidth halving mechanism worked with half duplex radios sharing a common channel. If speed becomes an issue, and a wired backbone is impossible, then you can always add another access point at each existing access point. With WDS, everyone has to use the same channel. However, with back to back repeaters, each pair of radios can use different non-overlapping channels and therefore can run full duplex. This eliminates the half duplex bandwidth reduction problem. If the Belkin boxes are really $35/ea, this might be a better idea.

Let me try again. I haven't seen your topology diagrams so I'm working from your description. As I understand it, you have one Belkin router connected to the internet. I'll call this the main wireless router. You also have 3 other Belkin AP's, each connected via WDS to the main wireless router. You MIGHT have those three Belkin AP's configured to also talk to each other. This is called a "star" topology in that each of the 3 AP's on the "spokes" of the main wireless router is exactly one wireless hop from the main wireless router. Therefore, you will see absolutely no bandwidth reduction from the main access point.

Now, it gets messy. If you want to send files between two of the PC's, and you have the 3 AP's setup to use WDS to talk to each other, then you will also see no reduction in bandwidth because the traffic is NOT going through the main wireless router. However, if you do NOT have the 3 AP's setup to use WDS to talk to each other, all such PC to PC traffic will go through the main wireless router via WDS, and you will see a reduction to 1/4 of the speed (two hops).

Your original question seemed to imply that you had 5ea WDS bridges connected in series, where each one passes data to the next. I declared this an abomination and based most of my comments on this topology. Methinks the star arrangement will be adequate for the task, especially if the PC's are plugged directly into the 3ea AP's. However, if you connect via a wireless laptop, you'll instantly pickup a 50% bandwidth reduction due to the half duplex, store and forward mechanism previous described in excruciating detail.

Hopefully, you mean Kbits as in kilobits. Units of measure are important. That appears to be almost exactly 1/4 of the 3Mbits/sec speed, which is what I would expect from a two hop WDS relay. That doesn't sound correct for a star topology.

That's odd as in something is wrong. If it's really a star type of topology, then the preformance should be almost identical at all points. What would be helpful is a better description of what you have plugged into what, and which AP is configured to talk to which other AP.

My wild guess(tm) based on insufficient information is that you have not set the WDS bridges MAC addresses properly so that they are talking to each other in a star type arrangement, where the centrally located main wireless router has the DSL line plugged in. That will cause the packets to go un-necessarily through excessive WDS hops to arrive at their destination, causing the observed bandwidth reduction.

Can you get 3000Kbits/sec measured at the main central wireless router with a local LAN connected (wired) computah?

I couldn't view the DXF file you posted. Something wrong with my ancient viewer. I just found a free DXF viewer and will try it later. In the future, JPG PNG or GIF please.

formatting link

I beg to differ. You have no less than 5 radios, all on the same channel, and all spewing what I consider to be excessive transmissions just to move one packet. That's what I consider to be the major problem with WDS, repeaters, and mesh networks. In order to move one lousy packet, they have to transmit that packet over and over again for each WDS hop. If your radio is in a position to hear all the traffic from your other AP's, there's going to be interfernce from your own system. If your topology had been the original 5 AP's in series, you would have 4 packets transmitted for every one received at the endpoints. All 4 packets are considered interference as none of them are needed by the endpoint. The good news is that your star type topology reduces the excessive packets to zero for PC's directly connected to the spoke AP's, and one extra packet to laptops connected via wireless. That's tolerable.

Good luck.

Reply to
Jeff Liebermann

I probably should have looked at the drawing before posting. It's NOT a star topology. It looks like this:

All boxes are Belkin 54G wireless access points with WDS enabled.

[ Cable ] [Main #1] W [ #2 ] W [ #3 ] [ Modem ]====[router ]==D==[ WAP ]==D==[ WAP ] | S | S | | | | | | | | | | wired LAN [ PC #1 ] [ PC #2 ] [ PC #3 ]

Speeds 2933 580 640 kbits/sec kbits/sec kbits/sec

Ok, this is NOT a star. It's a bus. The first directly connected router will get the full speed of the system. That's working.

#2 WAP is acting as a wirless client via WDS. There's no store and forward involved. It should be getting the full 2900kbit/sec. It's not. Something is wrong.

#3 WAP is acting like a wireless client via WDS from WAP #2 which is acting as a store and forward WDS repeater. There should be a 50% drop in bandwidth to WAP #3. Therefore it should be getting

1500Kbits/sec. Something (else) is wrong.
Reply to
Jeff Liebermann

Are you talking about niche ethernet cards that use the existing telephone or power wiring in the house or business to connect from point-to-point without having to install dedicated wiring? The ones that multiplex the signal on top of the existing telephone signal or the nominally ambient 60 Hz tone used by AC power lines?

I've tried both types I mention, and had bad experiences with both. I have NOT tried an alternative method using existing cable coax, and what you provide a link to below I'll look into.

You mean for best performance I shouldn't set them all up on the same channel, as I do have them now on channel 1?

Gotcha.

My original plan to set the five routers in a linear series to cross this building was based on finding places to hide the routers within 80-100 feet of each other, leaving the five routers in some sort of *rough* line diagonally across the building.

On that assumption, I also assumed WAP#2 wouldn't be able to talk to WAP#4, and therefore planned on only including the immediate adjacent WAP into the bridging table on each unit.

From what you tell me, I'll be much better off fully populating the bridging tables, even if I suspect I'm entering a MAC address for an unreachable WAP. After all, there is no penalty for a MAC entry that can't be reached.

Nearly 3.0 Mbits/sec at WAP#1 to my wired machine, WHEN CHARTER IS UNDER A LIGHT LOAD. During the daytime, betwixt 1.5 and 2.0 Mbits/sec.

Could you not view the WMF (Windows Meta File, a Windows vector drawing file type) I attached to that same message?

KHAN! ARGH!

I neglected to consider those radios.

I was only counting wireless telephones and neighbor's WAP's, of which there are none.

JEFF, THANKS VERY MUCH FOR YOUR HELP!

Reply to
Bill Evans

Jeff,

I'm going to try fully populating those MAC addresses into the bridging tables, and then tell you what happens.

Once again, thanks!

Reply to
Bill Evans

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.