Optical/IP Networks

US Data Services Are Awful

NEW ORLEANS -- CTIA Wireless 2003 -- Agilent Technologies Inc. (NYSE: A) today showed that wireless data service is universally horrible on currently available networks in the U.S.

Following in the footsteps of a similar test in Cannes last month (see Wireless Nets Suck, Says Expert), Agilent is running data transfer tests on the Code-Division Multiple Access (CDMA) 2000 1xRTT networks being operated by Sprint PCS (NYSE: PCS) and Verizon Wireless, and the general packet radio service (GPRS) networks being offered by AT&T Wireless Services Inc. (NYSE: AWE) and T-Mobile USA here at the show. Agilent's verdict: The data services on these networks need a serious amount of work.

The issue, according to Rice Williams, wireless network test division marketing manager at Agilent, is that carrier engineers aren't used to combining radio and IP networks in the way they now must, in order to create and manage "always-on" wireless data networks.

This is a big problem for the carriers, explains Rice. "IP is not one of their core competencies," he says.

Sadly, Agilent won't reveal which network is the absolute worst, even after many deep and probing questions from Unstrung (oh, alright... we offered them $20...).

Pundits may have expected CDMA 1xRTT networks to handle data better than GPRS networks. After all, they do offer faster data transfer rates -- around 40-50 kbit/s compared to the 20 kbit/s or so offered on GPRS.

However, Rafael Andrade, senior RF engineer for Agilent says that's only part of the picture. Once they start transmitting the data, the CDMA networks are faster, but first a session has to be set up between the network and the device, and that can add significantly to the amount of time a user waits to view a Webpage or download email.

At the show, Andrade had recorded session setup times of up to 50 seconds (!) before devices started to transmit data.

At the moment, users just can't expect to get the kind of quality of service they do on voice calls with data, Andrade reckons [ed. note: and let's face it, voice can be pretty ropey at times]. "It depends on so many different factors: what city you're in, how close you are to the transmitter… it's really not predictable.".

— Dan Jones, Senior Editor, Unstrung

Page 1 / 2   >   >>
lrmobile_castro 12/5/2012 | 12:25:04 AM
re: US Data Services Are Awful Not good. Not good at all. So much for the enterprise quality services.
hillcast1 12/5/2012 | 12:24:59 AM
re: US Data Services Are Awful I agree. From a developer's standpoint (streaming real-time quotes on PDAs and java phones), this makes me wanna cry. It seems that GPRS right now is one big cluster you-know-what.
vincentm 12/5/2012 | 12:24:48 AM
re: US Data Services Are Awful I don't know where you got that GPRS would offer 20 kpbs only ... most GPRS mobiles now support 4 timeslots, that means 40-45 kbps. When EDGE is deployed (and that should come pretty soon in the US), that will tripple the speed, reaching the standard ADSL throughput of 128 kbps over the air. Further, we Alcatel are still absent from the US in terms of radio infrastructure; that may be the reason why they observed such poor performances ! :-)
joset01 12/5/2012 | 12:24:43 AM
re: US Data Services Are Awful Actually Agilent ran this test in Europe last month and the results weren't exactly stellar there either (see http://www.unstrung.com/docume....

DJ Unstrung
jamo 12/5/2012 | 12:24:41 AM
re: US Data Services Are Awful Does anyone know what performance indicators were used and how they were measured?

FastFourier 12/5/2012 | 12:24:40 AM
re: US Data Services Are Awful For metropolitan area networks all of these providers need to start looking further down the road _if_they_want_to_solve_the_problem.

802.16a is new as a standard, and pretty clearly wide open for being the metric ofsuccess in solving this.

When people are standing on each others shoelaces (as they are now - just look at this thread), rather than each others shoulders then what they need is a goal sufficiently well described, and far away to reach for. With that we will get an orderly solution to this problem.

I think 802.16a is that goal.

The industry will nod in agreement..._if_they_want_to_solve_the_problem.

I just wish they had the courage to acknowledge the problem.
lrmobile_castro 12/5/2012 | 12:24:38 AM
re: US Data Services Are Awful 802.16 is (will be) a good solution for fixed wireless data but is it really optimized for the mobile space (yes even 802.16e)?

Or are you suggesting using 802.16 for nomadic-low speed.

I think the wireless world is just starting to reveal how difficult it is to do data over wireless at high speeds or even low speeds when service quality and low latency are required.

The mobile industry has a lot of work to do.
FastFourier 12/5/2012 | 12:24:30 AM
re: US Data Services Are Awful The 802.16a prototypes ( there being no certifying testbed yet for 802.16a) that are out there do offer high data rate "tests" at highway speeds ( and in one case to/from an airborne light aircraft). Both of these were in the ofdm ( not ofdma) PHY layer of the 802.16a triangle of PHY choices (ofdm,ofdma,singlecarrier)

see http://www.wi-lan.com/news/pre...
30Mbps @ 70 mph -- the Test-bed Center for Interoperability, an R&D Center operated by the California State Department of Transportation (Caltrans) hosted the demonstration. This was 2+ years ago.

It would be interesting to see the same tests again done with current (post 802.16a definition) technology. I don't know how the W-ofdm of 2000 differs from the W-ofdm of today.
vgholkar 12/5/2012 | 12:24:18 AM
re: US Data Services Are Awful I would not discount Agilent's tests. The empirical results emphasise that GPRS speeds are a function of several variables.

These variables include

(1) The maximum number of GPRS users that may use the network [capacity a big deal and is a function of numerous network and environmental factors in its own right, to get a feel check out books like Mobile Radio Propagation Channel by Parsons, Radio Interface System Planning for GSM/GPRS/UMTS by J. Lempi+ñinen & M. Manninen.

(2) How many device are on the network. [this is about congestion as opposed to capacity]

(3) Latency [where are the network bottle necks?]

(4) Coding scheme levels (there are 4 levels and these offer a tade-off between throughput and
coding protecion) - this also should remind us that protocol headers take up space so that 40kbps is not 40 kbps of data but 20kbps of data and header info. [where i define data to be info that we are after]

and so on.

And don't forget that download and upload times are different. If you have tried to send photos from your phone via email or mms using a gprs bearer you will know what I mean.


vgholkar 12/5/2012 | 12:24:17 AM
re: US Data Services Are Awful Correction:

(4) Coding scheme levels (there are 4 levels and these offer a tade-off between throughput and
coding protecion) - this also should remind us that protocol headers take up space so that 40kbps is not 40 kbps of data but 40kbps of data and header info. [where i define data to be the info that we are after].
Page 1 / 2   >   >>
Sign In