Slow Data Transfers?

Chuck Forsberg WA7KGX N2469R wrote:
> A theoretical possibility, given use of a cable adapter of some sort
> (USB female to USB female).

USB (Universal Serial Bus) is anything but universal. At least with the
old serial method, you could connect a device and equipment in any
combination using a null modem; where with USB one has to be the
"server" and the other the "client"

PC to phone, PC to web cam, and PC to printer works, but phone to web
cam or phone to printer won't work . Nor will PC to PC work
(server/server) unless if you have a smart cable that is a
"client/server" adapter (less than $50). There is no 'client/client"
adapter cable (female to female) made.

---

› See More: Slow Data Transfers?

[POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]

In <[email protected]> on Sat, 29 Apr 2006 02:06:11
-0700, Chuck Forsberg WA7KGX N2469R <[email protected]> wrote:

>On Wed, 26 Apr 2006 17:26:37 +0000, John Navas wrote:

>> How about transfer over USB cable?

>A theoretical possibility, given use of a cable adapter of some sort
>(USB female to USB female).

What you want is called USB On-The-Go (OTG).
<http://www.usb.org/developers/onthego/>

>I am taking the 6515 back. It's close but no cigar.
>Perhaps I am spoiled by the obsolete Kyocera 7135, but
>I expect a $$$ PDA/phone to come with tools to install
>my own ringtones. I expect an expensive phone to come
>with voice dial. I expect to answer calls with one hand
>without having to look at the phone. I expect to hear the
>other party with normal settings. In this day and age I expect
>the GPS to figure out where it is cold/warm start) without
>having to place a data call. And I expect a data phone
>to smoothly transfer data at close to the advertised speed.

Most "GPS phones" are actually A-GPS (Assisted GPS) -- to keep cost down and
to enable indoor operation, they require support from a base station.
<http://www.radio-electronics.com/info/cellulartelecomms/location_services/assisted_gps.php>

>I see uploading digital pics to one's home system as
>a "killer app". Such a phone could be designed today
>but there is no guarantee it will.

If it really is a "killer app" then it will undoubtedly emerge, since it's
pretty obvious. There are, however, a number of serious obstacles to
overcome, including simple and reliable device connection, and necessary
bandwidth (since high res digital pictures are huge). Even at an advanced 3G
speed of 400 Kbps, it would take about 1-1/2 minutes to upload a single 4 MB
picture, or well over an hour to upload just 50 pictures, which I think is too
slow for most consumers. Then there's the cost issue.

I think it more likely that we'll see this first with cell phones with lower
res integrated cameras, which are finally getting to the point of taking
decent pictures.

The vendor best positioned to do what you want is Sony, since it has big
stakes in both digital cameras and cell phones.

--
Best regards, SEE THE FAQ FOR CINGULAR WIRELESS AT
John Navas <http://en.wikibooks.org/wiki/Cingular_Wireless_FAQ>

[POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]

In <[email protected]> on Fri, 28 Apr 2006 17:10:43 -0400,
Isaiah Beard <[email protected]> wrote:

>wild bill wrote:
>
>> I notice latency is running around 1.4 seconds.
>>
>> (Hmmm. last week, in Kansas City, I think I saw
>> 800-900ms, still, an awfully long time.)
>
>Hmm, those ARE really long ping times, so I'm not sure what's going on
>there. The 500-600ms range though, is pretty much the norm.

Actual test results for Cingular EGPRS(EDGE) using Windows XP 'ping':

Packets: Sent = 20, Received = 20, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 384ms, Maximum = 642ms, Average = 532ms

>> Not sure what explains that - most WiFi's I've
>> used are well under 200ms. Does it have to
>> take that long?
>
>Well, yes. A WiFi access point is generally only a couple hundred feet
>away at most and might typically have maybe 20 or so users, max. Most
>cell sites are anywhere from half a mile to two or three miles away, and
>can have several dozens or even tens of dozens of voice AND data users
>vying for the same bandwidth.

The big difference is actually due to greatly different wireless protocols --
GPRS/EGPRS(EDGE) was designed to optimize network efficiency at the expense of
latency.

>> It's almost as bad as off an
>> earth orbiting satellite.
>
>Actually, an Earth Orbiting Satellite will typically offer ping times of
>2000ms.

Geosat latency is actually in the range of 240-280 ms for a single hop, or
480-560 ms round trip. It can be much worse than that when there is more than
one hop, and additional latency comes from the landline connection.

--
Best regards, SEE THE FAQ FOR CINGULAR WIRELESS AT
John Navas <http://en.wikibooks.org/wiki/Cingular_Wireless_FAQ>

[POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]

In <[email protected]> on Thu, 27 Apr 2006 14:39:44 -0500,
clifto <[email protected]> wrote:

>John Navas wrote:
>> clifto <[email protected]> wrote:
>>>John Navas wrote:
>>>> My current favorite (thanks, Jeff): <http://netspeed.stanford.edu/>
>>>
>>>It looks keen, but I routinely get more than the reported 2.32Mbps.
>>>Also seems odd they'd call the difference between my cable modem's
>>>upload and download speeds a "serious problem"; that's going to happen
>>>with any cable or ADSL connection.
>>
>> It's far from perfect -- I know of other issues -- but it's nonetheless a
>> useful tool.
>
>Agreed. The detailed report gives oodles of good information, almost too
>much (if there's such a thing). It did tell me my speed may be limited
>by a 128K comm buffer, so I'm searching through linux trying to find what
>allocates that and up the size a bit.

The comm buffer report is one of the things that not accurate. 128K is almost
certainly more than enough unless you have unusually high latency.

--
Best regards, SEE THE FAQ FOR CINGULAR WIRELESS AT
John Navas <http://en.wikibooks.org/wiki/Cingular_Wireless_FAQ>

On Fri, 28 Apr 2006 17:10:43 -0400, Isaiah Beard
<[email protected]> wrote:

>wild bill wrote:

>> Not sure what explains that - most WiFi's I've
>> used are well under 200ms. Does it have to
>> take that long?
>
>Well, yes. A WiFi access point is generally only a couple hundred feet
>away at most and might typically have maybe 20 or so users, max. Most
>cell sites are anywhere from half a mile to two or three miles away, and
>can have several dozens or even tens of dozens of voice AND data users
>vying for the same bandwidth.

re: latencies on cellular (internet) data links

at those distances (say 100 feet vs 5 miles), the difference will be
around 52 microseconds, round trip. I doubt most earthlings have
any sensory capability to detect that directly.

(I used a wild-ass approximation of 1ns per foot, and 26,000 feet
one way. Get the 'right' numbers and calculate it to a gnat's ass
if you want. I just wanted the ballpark, and order of magnitude)

Cell sites use a very sophisticated supervisory system to keep
collisions & crosstalk to a minimum, along with hundreds of active
channels available at any one time. WiFi does not.

Actually, the worst 'crosstalk' (well, it sounded like crosstalk) I've
ever experienced was in Manhattan, where maybe as many as six
or seven T-Mobile cell sites were all 'visible' to my phone at the
same time. Maybe it got confused?

Then there's cable TV. Ever check out that activity light on
your cable modem? For that matter, anybody got a way to
packet sniff the cable? Or, set a promiscuous 'listen only'
mode in their modem?

It's not distance, and it's not the number of users.

Measure it, tell us how you measured it, and what you saw.

I've seen 1100 milliseconds (+/- two or three hundred),
using PingPlotter, T-Mobile, and a Sierra 775. You?

Bill

On Tue, 25 Apr 2006 01:37:41 -0700, randy wrote:
> 56k bits (not bytes) /sec, no pop3, no news server,no web space,
>
> $60. plus a month I'm really rethinking it my self.

Sounds like a really bad EDGE connection. For my $60/month I get ~300kbs,
but that's using 3G UTMS. Latency is horrible, and TCP/IP connection
establishment abysmal. But, it's tolerable, at least until HSDPA debuts in
handsets. HSDPA is a UMTS upgrade advertised as "BroadBand Connect" by
Cingular (most Cingular reps have probably never heard of HSDPA), but
currently only supported in pc-card CDMA chipsets.

For handsets, the only ones supporting UMTS are the LG CU320 and Samsung
ZX10, and of those only the LG supports Bluetooth 3G "DUN". (Tell that to
the reps at my local Cingular store; they were telling customers the
Samsung ZX10 did bluetooth till I made them read the specs on the box.)

The Samsung ZX20 will be the first handset which does UMTS+HSDPA, and I'm
eagerly awaiting it since HSDPA is _supposed_ to be the answer for poor
latency with UMTS, which currently is around 300-600ms just for reaching
the Cingular network gateway in Atlanta.

FWIW, this is in the San Francisco Bay Area. I can vouch for parts of San
Francisco, Mountain View, Sunnyvale and San Jose.

[POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]

In <[email protected]> on Mon, 08 May 2006
17:34:29 -0700, William Ahern <[email protected]> wrote:

>On Tue, 25 Apr 2006 01:37:41 -0700, randy wrote:
>> 56k bits (not bytes) /sec, no pop3, no news server,no web space,
>>
>> $60. plus a month I'm really rethinking it my self.
>
>Sounds like a really bad EDGE connection. For my $60/month I get ~300kbs,
>but that's using 3G UTMS. Latency is horrible, and TCP/IP connection
>establishment abysmal.

Sounds like something may be wrong -- TCP/IP connections aren't an issue for
me even on EGPRS(EDGE).

>But, it's tolerable, at least until HSDPA debuts in
>handsets.

HSDPA as compared to UMTS is of relatively little value in handsets alone.

>HSDPA is a UMTS upgrade advertised as "BroadBand Connect" by
>Cingular (most Cingular reps have probably never heard of HSDPA), but
>currently only supported in pc-card CDMA chipsets.

Actually *W*CDMA, which is quite different from what's commonly called "CDMA",
and backward compatible with GSM EGPRS(EDGE) and GPRS.

>For handsets, the only ones supporting UMTS are the LG CU320 and Samsung
>ZX10, and of those only the LG supports Bluetooth 3G "DUN". (Tell that to
>the reps at my local Cingular store; they were telling customers the
>Samsung ZX10 did bluetooth till I made them read the specs on the box.)
>
>The Samsung ZX20 will be the first handset which does UMTS+HSDPA, and I'm
>eagerly awaiting it since HSDPA is _supposed_ to be the answer for poor
>latency with UMTS, which currently is around 300-600ms just for reaching
>the Cingular network gateway in Atlanta.

Except for interactive things like SSH, I don't find the latency to be a big
issue, particularly for the handset alone. For a notebook computer my
personal preference is to use a PC Card. For HSDPA that would currently be
the Novatel U730.

--
Best regards, SEE THE FAQ FOR CINGULAR WIRELESS AT
John Navas <http://en.wikibooks.org/wiki/Cingular_Wireless_FAQ>

On Tue, 09 May 2006 05:40:17 +0000, John Navas wrote:
> [POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]
>
> In <[email protected]> on Mon, 08 May 2006
> 17:34:29 -0700, William Ahern <[email protected]> wrote:
>
>>On Tue, 25 Apr 2006 01:37:41 -0700, randy wrote:
>>> 56k bits (not bytes) /sec, no pop3, no news server,no web space,
>>>
>>> $60. plus a month I'm really rethinking it my self.
>>
>>Sounds like a really bad EDGE connection. For my $60/month I get ~300kbs,
>>but that's using 3G UTMS. Latency is horrible, and TCP/IP connection
>>establishment abysmal.
>
> Sounds like something may be wrong -- TCP/IP connections aren't an issue for
> me even on EGPRS(EDGE).
>
>>But, it's tolerable, at least until HSDPA debuts in
>>handsets.
>
> HSDPA as compared to UMTS is of relatively little value in handsets alone.

I should've mentioned that I tether. Also, I prefer to use my handset,
rather than a pc-card. Especially since UMTS can do voice and data
simultaneously. My iBook can't even take a pc-card, anyhow.

>>HSDPA is a UMTS upgrade advertised as "BroadBand Connect" by
>>Cingular (most Cingular reps have probably never heard of HSDPA), but
>>currently only supported in pc-card CDMA chipsets.
>
> Actually *W*CDMA, which is quite different from what's commonly called "CDMA",
> and backward compatible with GSM EGPRS(EDGE) and GPRS.

Yep. Sorry

>>For handsets, the only ones supporting UMTS are the LG CU320 and Samsung
>>ZX10, and of those only the LG supports Bluetooth 3G "DUN". (Tell that to
>>the reps at my local Cingular store; they were telling customers the
>>Samsung ZX10 did bluetooth till I made them read the specs on the box.)
>>
>>The Samsung ZX20 will be the first handset which does UMTS+HSDPA, and I'm
>>eagerly awaiting it since HSDPA is _supposed_ to be the answer for poor
>>latency with UMTS, which currently is around 300-600ms just for reaching
>>the Cingular network gateway in Atlanta.
>
> Except for interactive things like SSH, I don't find the latency to be a big
> issue, particularly for the handset alone. For a notebook computer my
> personal preference is to use a PC Card. For HSDPA that would currently be
> the Novatel U730.

Yes, SSH can be painful. I dunno why, but connection establishment can
take forever in general. And it really shows on graphics intensive web
pages (and I assume when HTTP pipelining isn't being used). All those HTTP
connections, which can take 1-2 seconds to setup before even beginning to
download can really hurt.

Of course, I've had other troubles as well, like an inability to access
certain, usually west-coast, servers, but only when connecting from San
Jose. Bizarre, but I see it every day. Symptom is being able to connect
and receive data (or at least one packet of data), but not being able to
send anything. So, for instance, with SSH I can get a banner, but can
never send anything. Similar behavior for HTTP or anything else (SMTP, any
random service I setup with netcat).

[POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]

In <[email protected]> on Tue, 09 May 2006
11:29:35 -0700, William Ahern <[email protected]> wrote:

>On Tue, 09 May 2006 05:40:17 +0000, John Navas wrote:

>> Except for interactive things like SSH, I don't find the latency to be a big
>> issue, particularly for the handset alone. For a notebook computer my
>> personal preference is to use a PC Card. For HSDPA that would currently be
>> the Novatel U730.
>
>Yes, SSH can be painful. I dunno why, but connection establishment can
>take forever in general. And it really shows on graphics intensive web
>pages (and I assume when HTTP pipelining isn't being used). All those HTTP
>connections, which can take 1-2 seconds to setup before even beginning to
>download can really hurt.

On the phone, or tethered PC?

>Of course, I've had other troubles as well, like an inability to access
>certain, usually west-coast, servers, but only when connecting from San
>Jose. Bizarre, but I see it every day. Symptom is being able to connect
>and receive data (or at least one packet of data), but not being able to
>send anything. So, for instance, with SSH I can get a banner, but can
>never send anything. Similar behavior for HTTP or anything else (SMTP, any
>random service I setup with netcat).

Any links I can test? I'd like to see what I get. Thanks.

--
Best regards, SEE THE FAQ FOR CINGULAR WIRELESS AT
John Navas <http://en.wikibooks.org/wiki/Cingular_Wireless_FAQ>

On Tue, 09 May 2006 18:36:47 +0000, John Navas wrote:
>>Of course, I've had other troubles as well, like an inability to access
>>certain, usually west-coast, servers, but only when connecting from San
>>Jose. Bizarre, but I see it every day. Symptom is being able to connect
>>and receive data (or at least one packet of data), but not being able to
>>send anything. So, for instance, with SSH I can get a banner, but can
>>never send anything. Similar behavior for HTTP or anything else (SMTP, any
>>random service I setup with netcat).
>
> Any links I can test? I'd like to see what I get. Thanks.

Of course, last night these worked just fine. But these two are typically
rather consistent. More often than not I run into trouble.

69.80.195.225
64.62.167.198

There have been other, random web sites. The only consistent thread were
their west-coast'ish locations.

When the above addresses are inaccessible I typically tunnel through
some machines in Atlanta or DC to get access.

[POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]

In <[email protected]> on Wed, 10 May 2006
10:57:31 -0700, William Ahern <[email protected]> wrote:

>On Tue, 09 May 2006 18:36:47 +0000, John Navas wrote:
>>>Of course, I've had other troubles as well, like an inability to access
>>>certain, usually west-coast, servers, but only when connecting from San
>>>Jose. Bizarre, but I see it every day. Symptom is being able to connect
>>>and receive data (or at least one packet of data), but not being able to
>>>send anything. So, for instance, with SSH I can get a banner, but can
>>>never send anything. Similar behavior for HTTP or anything else (SMTP, any
>>>random service I setup with netcat).
>>
>> Any links I can test? I'd like to see what I get. Thanks.
>
>Of course, last night these worked just fine. But these two are typically
>rather consistent. More often than not I run into trouble.
>
>69.80.195.225
>64.62.167.198
>
>There have been other, random web sites. The only consistent thread were
>their west-coast'ish locations.
>
>When the above addresses are inaccessible I typically tunnel through
>some machines in Atlanta or DC to get access.

What service(s) can I test?

In the meantime, I did some experimenting, and was able to get this working
connection:

>telnet 64.62.167.198 POP3

+OK POP3 wilbur.25thandclement.com 2004.89 server ready
QUIT
+OK Sayonara

Connection to host lost.

I'm currently located in Alameda (on San Francisco Bay), connecting over MEdia
Net EGPRS(EDGE) with either:

* Bluetooth connection to a Cingular-branded Motorola V551.
* Sony Ericsson GC82 PC Card

--
Best regards, SEE THE FAQ FOR CINGULAR WIRELESS AT
John Navas <http://en.wikibooks.org/wiki/Cingular_Wireless_FAQ>

On Fri, 12 May 2006 19:34:30 +0000, John Navas wrote:
>>> Any links I can test? I'd like to see what I get. Thanks.
>>
>>Of course, last night these worked just fine. But these two are typically
>>rather consistent. More often than not I run into trouble.
>>
>>69.80.195.225
>>64.62.167.198
>>
>>There have been other, random web sites. The only consistent thread were
>>their west-coast'ish locations.
>>
>>When the above addresses are inaccessible I typically tunnel through
>>some machines in Atlanta or DC to get access.
>
> What service(s) can I test?
>

Any of them. 110, 134, 995, 993, 80, 443, 22, 563, 25.

The behavior was always identical (SYN, SYN+ACK okay, and at least one
packet downstream, then nothing), and likely IP related routing issues,
not application filtering.

Ironically, for the past few days things have been okay. The longest
stretch yet. I guess I just needed to voice my issues aloud