Range Extenders ... Is "Conventional Wisdom" Correct?

[stevech]

A WiFi repeater may receive frames at rate A but transmit them at rate B. These will differ according to the signal to noise and interference conditions on each of the two links.
One example: weak signal on client device link, strong signal on link to WiFi router. So the latter uses less air time for the same amount of data.

(this assumes a single-radio repeater).

 

[Klueless]

Still messing around with benchmarking. I chose netstress because it looked simple but my results looked more like it was from 100 Mbps Ethernet rather than Gigabit. Finally got a chance to look a little closer. Turns out both PCs are Gigabit capable. Then I assumed house wiring might be only CAT 5 so I layed out some CAT 5E. Results still looked like 100 Mbps Ethernet. Netstress seemed to give me the same max packet size whether I was set for 100 or 1,000 Mbps so I upped data streams from 1 to 8. Finally got some results that looked like something one might get from Gigabit Ethernet (as compared to the ones a few messages back).

Hopefully I'll get a chance to play more next week and fine tune my methodology. But, until later, I'm beginning to understand what great tools these are for fine-tuning the positioning of a range extender (or possibly an AP). I was also finally able to answer my questions as to why I was getting better throughput through a range extender than going direct.

Wi-Fi analyzers are great but in the end it's really all about the data. My analyzer helped me pick out what channel to use and where to place my range extender. For 2.4Ghz it suggested the front side window and for 5Ghz it preferred I use the 2nd window. As I ran through my test suite I noticed huge increases in performance in the main bldg. and substantial increases in bldg. 2. Again I wondered why I got better performance through the range extender versus connecting my laptop to the same SSID in the main bldg. that my range extender used.

On a whim I connected my laptop directly to the Ethernet port on my range extender (an inside out AP if you will : -) Bam! 150Mbps ... about double what I was seeing though the range extender ... just like everyone has been telling me!

Then I noticed that my range extender had shifted out of position a little so I moved it back an inch or two (yeah, I might be a little OCD) and I noticed netstress was now reading 180. Moved it back and forth a couple of times and netstress moved from 150-ish to 180-ish each time. Then I fiddled with the antennae a little and quit at 213.

So between moving my extender 2 inches and changing my two antennae from || to _| I picked up an extra 60Mbps in performance. So even though I shot my laptop through the same window to the same router it's no longer a mystery why going through the range extender was faster. I was comparing Apples to Oranges. If moving my range extender two inches can have such a profound effect then sitting at my laptop three feet down and a foot over could most certainly account for the discrepancy I was seeing.

My other takeaway (probably obvious to you vets) is to test actual throughput before final placement of your range extender or AP. After all, it's all about the data ... isn't it?

 

[sfx2000]

My other takeaway (probably obvious to you vets) is to test throughput before final placement of your range extender or AP.

With WiFi, whether AP or Repeater, location is everything...

 

[Klueless]

location is everything...

We sound like realtors : -)

 

[sfx2000]

We sound like realtors : -)

Yep - and moving a client or AP sometimes as little as 6 inches can make all the difference...

http://arstechnica.com/gadgets/2014/08/mapping-wi-fi-dead-zones-with-physics-and-gifs/

 

[Klueless]

... as little as 6 inches can make all the difference

Love the chart.

At 2.4Ghz window one was the right choice. At 5Ghz window two was the better choice. Interesting side effect though. We're now more sensitive to cars parking between the windows. I'm thinking of putting a sign by the one window, "Reserved Parking for Owner Only". (Should work, he's never there : -)

 

[sfx2000]

I'm thinking of putting a sign by the one window, "Reserved Parking for Owner Only". (Should work, he's never there : -)

Well - perhaps mark it "Customer Parking" and he might be there more often

Anyways - somehow I'm getting the drift that you're much less clue-less that your handle indicates - nice work on your investigation and problem solving...

Well done!

 

[Klueless]

nice work ...

Thank you kindly, it's been a long three months here at the University of SNB : -) Here's a shot of the EX6150 (upper right corner) pointed at the window across the way (left of red sign). You will probably need to login to view.

When the network slows down I can pretty much guess the boss just pulled in with his humongous Cadillac Escalade. (On the upside I always know when break time is over : -)

Playing with my benchmarks it would seem Netstress is either showing its age and/or has a bug. It defaults to 100Mbps Ethernet (age?) and, as such, sends up to 8,127 packets per second (PPS). So, as per instructions, I set myself to Gigabit Ethernet and PPS to 81,272. Netstress remembers the Gigabit part but keeps dropping PPS back to 8,127 (bug?).

So instead of running only one data stream I compensate by running 8 TCP streams and 8 UDP streams concurrently. Not perfect but at least I'm getting what I need. (I reread an article by Mr. Higgins, apparently he ran into the same thing.)

 

[Klueless]

We've all seen it, "Use a wired AP. When you use a range extender your times will be halved."

Just wanted to take a few minutes and thank everyone for their input. I learned a lot and I thank you all for that!

The original question was, Boo, range extenders, evil. They'll double your times, half your throughput. Is conventional wisdom correct?

And the answer is yes, but ...

At my son's large (by my standards) family home Wi-Fi runs amazingly well. A house full of kids and everyone is happy. Except for the master bedroom. It could not be further from the router if you had designed it that way. We could have run cable, well not us, we would have had to hire a professional, a few hundred dollars, but it could have been done.

But for fifty bucks we picked up a cheap range extender, about the size of a pack of cigarettes, plugged it into a wall outlet and five minutes later his wife was streaming movies and catching up on Facebook from the comfort of her bed.

Or, back at the car lot, I simply couldn't. I'm 70 years old. I wasn't about to grab a pick ax and shovel and break through a hundred feet of black topped parking lot to lay a cable. All said and done it would have been about five grand to an outside contracter.

I had a contractor come in who proposed an Ethernet radio bridge but the owner turned that down too. I work for a used car dealer. It just ... wasn't ... going ... to happen. So I did what I did, I used a range extender.

In any case let's talk about the data. In bldg. 2 a couple users could connect wireless to the main bldg. Kinda. Slow, intermittent, lots of disconnects. The rest couldn't.

We picked out the best place to connect and set a wireless laptop there. We then fine tuned the placement of the range extender by mounting it on a shelf about three feet above the desk and another two feet over. Here's the data:

• 56 Mbps - Laptop wireless to main bldg.
• 93 Mbps - Laptop wireless to range extender (wireless to main bldg.)
• 215 Mbps - Laptop hardwired to range extender (wireless to main bldg.)

So yes, 93 compared to 215, throughput was definitely halved.

But 93 compared to my best case user with 56 ... definite improvement (which is what misled me to my original question). As I walk through bldg. 2 everything, every room, works, no drops. The guys who split time between the two buildings can't tell the difference. (Of course we've only got a 15 x 1.5 Mbps Internet service so we do keep expectations on the low side : -)

Now we've the original leg (leg A) from bldg. 2 to the main bldg. and the new leg (leg B) which is bldg. 2 users to the range extender. 2X. Or, maybe, a little less ...

• Leg B (range extender) should be a little faster than original leg A, that's why we put in the range extender to start with.
• Leg A should be faster than the original leg A
  • We dropped the deadwood.
  • We optimized positioning.
  • We have no bldg. 2 clients on the back haul radio
    • If we had a tri-band router we would have no clients on the back haul radio

Case in point, 93 Mbps compared to my best case user with 56. Intriguing, compared with my old reality I'm 65% better but ... compared to my new reality my speeds are indeed halved!

I do concur (my original premise was wrong) with the consensus (use wired APs) but;

• They're not evil, range extenders can (and do) work (for some applications)
• You can hard wire clients to the range extender (turning it into a kind of inside out AP ; -)
• Some range extenders can be converted to a wired AP later

I think I finally (almost) understand. Thanks again!

 

[sfx2000]

It takes a village sometimes... borrowing a phrase or two...

Sometimes it's just about getting coverage/connectivity - and to this end, you've found a solution.

Thanks for sharing, and please, don't be a stranger - your contributions back to the SNB forums here are appreciated.