I'm curious to know if anyone has played around with the TCP settings such as TCP Window Size, Default TTL, etc in order to "optimize" or tune the TCP stack for EVDO? I have done this before on Windows 98/2000/XP using a broadband connection with amazing results, but I have not seen any mention of doing it with WM for EVDO.

This can obviously be attempted through the Registry, but there is also one app I know of that claims to do it as well:

http://www.freewareppc.com/communication/tcptune.shtml

I would gladly experiment and run some trials with Verizon's EVDO here in Atlanta, but I would need some suggestions on the settings to use. I understand the concept of how these settings make a difference, but I don't know enough about EVDO networks to know what tweaks to make.

Has anyone attempted this, or does anyone have any helpful info? (Sorry if this is a repeat thread, I searched for quite a while without finding anything.)

TIA-

My concern would be that changing network loads would necessitate change in the window size. It might be difficult to find the "best" receive window.
However, I am far from an expert with this stuff.

If you do, go carefully and with fully functional backups of your registry and handset.

Good Luck,

Bob Duckworth

__________________
My cellphone is smarter than your honor student.

from what I can tell, rwin is already set at 65536 so going any higher might not help much. Going lower might be an interesting experiment though since evdo is such a high-latency connection.

While surfing around I came across the following page...anyone seen this??

http://evdo-coverage.com/pocket-pc-pda-ppc.html

Based on the outlandish claims they are making (not to mention the ridiculous $585 price tag), I'm guessing these guys are complete scam artists. However, lets assume for a second they are able to tune, optimize or boost EVDO speed even a small percentage of what they claim... how are they likely doing this? They apparently are using the TCPTune app I mentioned above, plus who knows what else.

Thoughts?

Interesting.. Seems kind of scam-ish though...

Agreed, seems very scam-ish. But I have to think that they are at least achieving SOME results in the bandwidth boosting dept. I mean, could they really claim an utterly ridiculous 1500Kb DL speed if they aren't even boosting it over the average ~500Kb that most users get out of the box? You'd think they are at least improving it slightly...

well, i just ran tcptune after doing some stock speedtests on dslreports.

My office is inside a metal walled/roofed warehouse, so signal strength is pretty much crap with whatever band wireless we are talking about, be it 800, 1.9, 4.9, 2.4, 2.6, 5.3, or 5.8

Anyway, stock speeds with the loading bay door open were about 500k average. Loading bay door down, average was 300k

Run tcptune, and door is down now, i get about a full meg down. Nice huh?

So yeah, just run it.

oh yeah, and that site is just a memory card with those apps on it. Do you really need someone to install those apps for you? If you can make it to this forum, then you should certainly be able to load software onto your device. It ain't hard, and there are plenty of people here willing to help for free if you run across issues.

bak,

so you downloaded the program tcptune and just ran it? did you do any tweaks inside the program itself or just left it alone?

Sort of like one of those cell phone "signal boosters" you stick to your battery

BakCompat, when you ran TCPtune, what settings did you use? I'm assuming that for Network Adapter you chose "Cellular Line". After that, can you elaborate on your choices? Did you change the TCP Receive window (which I'm assuming you would have to in order for this app to actually make a difference)? Did you use any "Optimize for" commands in the Settings menu?

I'm asking because I have tried several combinations now without seeing any difference. Hopefully we can collectively determine if this app can help us at all.

TIA

Whoa! I ran it and didn't make any changes. I got a string of better than 800 with a 925 thrown in at the dslreports site. I was averaging 450 or so back in December. I'm sold!

I'll try to get more scientific tomorrow with a couple coworker that have 6700s. (One of them is supposed to be setting up a test Sensa server...)

If you guys wonder about the integrity of those guys on the website look at this:
http://evdo-coverage.com/evdo-v650-l...od-router.html

They call the Kyocera KPC-650 the v650... they also state that the so called EVDO model is $350... Not true. It only has an EVDO model. And It's $179 with a 2 year, $229 with a 1 year, and it's $279 with no service contract.

I could not find there so called EVDO wireless router, the "MIS" Grade one anywhere online. And the so called Lightning Rod, reception booster, they claim even extends battery life on your Laptop!! How that's possible I'll never know.

They obviously don't know Verizon products and obviously don't know what they're talking about.

anyone else test this yet? is there an option to revert to original settings? if not, would that require a registry backup, complete backup? both? hard reset?

thanks,

-jeff

__________________
AT&T Tilt aka HTC Kaiser aka TyTn II
PPC 6700 is now my TV remote control and favorite paperweight lol

I decided to take the plunge! Why not, 8+ hard resets testing 3rd party apps with the 6700 & WM5

Here is what I have found with TcpTune v2.01:

In my area I only get about 0-2 bars relative to cellular radio freq. So my EDO was pretty pathetic.

Constantly no higher than 200 kbit/sec, mostly 70-120. I played with TcpTune this morning and changed-

Network Adapter: Cellular Line
TCP Receive Window: 32768 (org was 65536)

All other settings left default.

I also increased the TCP Window in increments of 8 with no real improvements. I decided to cut the default value in half (32768 )

My tests using DSLreports now show my speeds to be 300-500 consistently never dropping below 200 this is after 2 hours of testing from the same location.

I also notice my EDO connection stays connected longer.

Another test to end this post.........398


I'm Sold!

__________________
Iron sharpens iron, so one man sharpens another. Proverbs 27:17

Marctronixx & topkat2000,

Now as i said before, my office is inside a "galvanized metal" type walled/roofed warehouse. We have several types of wireless gear here across about 8 or so different frequencies. Pretty much all of them are inhibited by the metal content to varying degrees. I would have expected a modest gain from tcp tuning, but i was quite surprised that i was able to double the effective download speed. The dslreports test is only a 600k file, so isn't as "accurate" when downloading at 1 meg. So, i ran some speedtests on our company website, and got similar results.. So yeah, i was able to confirm it for my particular experience.


I didn't do anything fancy, as using a preset configuration optimization was all that was necessary. Load the program, it's already got "cellular line" as the selected device to tune. TCP receive window was already specified at the value of 65536. I think i tried the optimizations for FOMA and PacketOne. I'd like to do some more testing with each optimization for comparison benchmarking, but probably will not be able to do so until this weekend. Being inside what is effectively a faraday cage really impacts on my wireless accuracy. Any results i get here will certainly not reflect upon yours.


Probably the single most important thing to change is the RWIN value. tcptune is nothing more than an interface to change the values for different TCP values. I suspect the optimizations will not have too much impact, but again, i'll need to do some testing when i'm not in a faraday cage.


Oh yeah, my average latency to dslreports was ~0.9 seconds.

Just thought i might add that i saw this thread and DLed TCPtune and ran it with default settings and now Agile Messenger won't connect. Anyone else having similar results? any idea what it changed that could be causeing this? Unfortunately I did not backup and i think i could be screwed

__________________
XV6800, Aliph Jawbone

I've been playing with this today also, but my Agile Messenger connects fine.

the 600k file should be fine for our tests - it is 600 kilobytes whereas we are downloading (hopefully) at 1megabit (1024 kilobits = 128 kilobytes).

-jeff

__________________
AT&T Tilt aka HTC Kaiser aka TyTn II
PPC 6700 is now my TV remote control and favorite paperweight lol

helix,

yeah same here. i just installed agile yesterday and hand not set it up yet. i just tried to set up one of my accounts on agile and it wont connect either... i then went to try to log on to pocket msn and could not get on.

i uninstalled tcp tune and those programs work. so tcp tune was the issue...

my test - 3 bars EVDO signal on Universty of Miami campus - south miami/coral gables area - all tests are being done in the same classroom without moving the phone from the desk within a 30 minute period of time.

before tcptune:

636k, 611k, 604k, 543k, 588k

after tcptune (no settings changed, no soft reset, install only...same place, same time, moments later after install):

635k, 683k, 719k, 714k
914k (after getting this speed on my 5th and final test, i decided to run 2 more to confirm it was an OUTLIER):
687k, 609k

after soft resetting and running the speed tests again:

597k, 594k, 577k, 687k, 717k

still testing...gonna play with rwin soon. so far i'd say just installing the program does not have any considerable effect on download speeds - i will speculate that the speed increase was just a result of natural network fluctuations...i'd like to see the average go up and the individual tests more consistently higher before saying there's a different...we'll see with rwin. i'll try the 50% rwin value first - anyone have any other rwin values i should try?

i continued testing speeds after typing the above paragraph - no changes and now i am noticing more bandwidth but it's got to be a result of the network fluctuations...which is making this test very difficult to interpret:

717k, 729k, 685k, 770k, 854k, 789k, 620k

ok - set my "cellular line" RWIN to 32768 (50% of original):

512k, 537k, 464k, 591k, 531k, 584k, 541k, 603k, 598k - worse, maybe no change

set my "cellular line" RWIN to 49152 (75% of original):

601k, 570k, 645k, 489k, 604k, 756k, 931k (?!), 763k, 916k, 736k
(notice the huge change in speeds from network fluctuations...all items within each set of tests are done back to back (mere seconds between tests))

set my "cellular line" RWIN back to 65536 (original value):

530k, 565k, 470k, 705k, 783k, 796k, 663k, 795k, 822k, 819k

uninstalled tcptune, soft reset, test results:

644k, 673k, 603k, 535k, 668k, 691k, 985k, 845k, 899k, 922k

why does it seem that when i do 10 tests, the last 5 are the fastest? having just gotten 4 tests near 900k i don't think there's a need for me to reinstall tcptune unless someone finds some optimal configuration changes to make - also, perhaps this test should also be run in an area that gets slower speeds (i get faster speeds here on campus than at my appartment so i'll redo this test when i'm there tonight and post the results)

i guess make your own interpretations...

-jeff

__________________
AT&T Tilt aka HTC Kaiser aka TyTn II
PPC 6700 is now my TV remote control and favorite paperweight lol