PDA

View Full Version : Why your carrier's reception sucks



tcn33
22nd February 2010, 06:16 PM
Seems the folks at AT&T and O2 have identified the reason their networks suck donkey balls - and it's the iPhone's fault.


Normally, devices that access the data network use an idling state that maintains the open data channel between the device and the network. However, to squeeze even more battery life from the iPhone, Apple configured the radio to simply drop the data connection as soon as any requested data is received. When the iPhone needs more data, it has to set up a new data connection.

The result is more efficient use of the battery, but it can cause problems with the signaling channels used to set up connections between a device and a cell node.

Interesting stuff, and the described symptoms ("frequent dropped calls, lack of voice mail notifications, inability to make or receive calls even when the signal looks strong, and inability to make data connections") may sound very familiar to some. Plus this at least gives some hope that there's a fix for the issue, aside from building a metric shit-ton of new towers.

How smartphones are bogging down some wireless carriers (http://arstechnica.com/gadgets/news/2010/02/how-smartphones-are-bogging-down-some-wireless-carriers.ars?utm_source=rss&utm_medium=rss&utm_campaign=rss)

bennyling
22nd February 2010, 07:02 PM
I could see how this makes sense from an Optus/other over-sold carrier point of view, but I'm pretty sure this is still a carrier issue - you've got to have a network to connect to in the first place, right?

Exocet
22nd February 2010, 07:37 PM
They're talking out of their ass. Voice calls and data sessions are two completely different things and a handset re-establishing a data connection has no bearing on the ability to make or receive voice calls, or SMS.

LennyX
22nd February 2010, 07:50 PM
They're talking out of their ass. Voice calls and data sessions are two completely different things and a handset re-establishing a data connection has no bearing on the ability to make or receive voice calls, or SMS.

Couldn't re-establishing hundreds of connections potentially overload the tower so much that it does affect voice calls?

Al Aero
22nd February 2010, 07:57 PM
Couldn't re-establishing hundreds of connections potentially overload the tower so much that it does affect voice calls?

Optus has ONLY one tower in every state.

Hence overload.

Bogus Jimmy
22nd February 2010, 08:00 PM
And this constant connecting and dropping only seems to tax the networks that were running close to capacity. That makes sense.

The difference between a good and a bad network is still night and day (used my iPhone on Optus for about a year and have been on Telstra for about a year).

mechcon
22nd February 2010, 08:00 PM
Wait, what? only one tower? you're kidding right? And folks, I'd like to confirm this with somebody else, but Vodafone use Optus' towers for 3G right?

Al Aero
22nd February 2010, 08:03 PM
Telstra's network blows this theory to shit..

I was on Optus for 18 months and it was nothing more that a complete disaster, no data, dropped calls, delayed messages etc etc.

Been on Telstra since December and have not had one issue.

LennyX
22nd February 2010, 08:06 PM
Wait, what? only one tower? you're kidding right? And folks, I'd like to confirm this with somebody else, but Vodafone use Optus' towers for 3G right?

Yeah they were joking. Optus have more than one tower but not enough to meet capacity.

And no, Vodafone don't use Optus's towers AFAIK. Vodafone gets 3G in my friends apartment when Optus only gets GPRS. It's Virgin (and a heap of other smaller telco's) that use Optus's towers.

Shaun R
22nd February 2010, 08:06 PM
Wait, what? only one tower? you're kidding right? And folks, I'd like to confirm this with somebody else, but Vodafone use Optus' towers for 3G right?

That was SARCASM!! :P

And wrong. Vodafone is an independent network.
Virgin uses Optus' 3G Network. Virgin is part of Optus.

Technically, they're all the same as each other. Don't quote me on this, but...

Telstra has a deal with Three, Three is part of Vodafone, Vodafone has a deal with Optus, Optus has a deal with Telstra, Telstra has a deal with Vodafone... Confused?

They're all as bad as each other!! :(

Back on topic, I think this is balls. Carriers just trying to get away with their crappy networks. If a phone is constantly MAINTAINING a connection, wouldn't that mean MORE load than if it only connects when NEEDED?!
This is balls, I say.

tcn33
22nd February 2010, 08:08 PM
Telstra's network blows this theory to shit.

Not necessarily - the theory just says it's an issue if the network isn't configured to handle "devices connecting and disconnecting at a much higher rate than they've been accustomed to".

the_doctor
22nd February 2010, 08:09 PM
Telstra's network blows this theory to shit..

I was on Optus for 18 months and it was nothing more that a complete disaster, no data, dropped calls, delayed messages etc etc.

Been on Telstra since December and have not had one issue.

Agreed I was also with Optus for nearly 2 years and I had constant call drop outs, electronic alien voices, constant call failures, delayed messages, no or extremely slow data, the lot. Since moving to Telstra I have not had one single issue either.

Al Aero
22nd February 2010, 08:12 PM
Not necessarily - the theory just says it's an issue if the network isn't configured to handle "devices connecting and disconnecting at a much higher rate than they've been accustomed to".

And thats Optus's problem in reality?

tcn33
22nd February 2010, 08:14 PM
^ *shrug* I'm not a network engineer. Apparently Ars' source is though, and the described symptoms sound pretty familiar.

timmytomtam
22nd February 2010, 08:18 PM
Just moved house and am waiting to get ADSL connected.

In theory, using PDAnet on the iphone would be fine for a few weeks, until we get broadband.

Oh how wrong I was. My sisters Telstra phone has full 3G reception everywhere in the house. My optus iPhone? Has to be sitting in the edge of the upstairs balcony to get 1 bar of flaky 3G reception.

Can't wait till August when I an ditch optus and move to Telstra :)

Exocet
22nd February 2010, 08:28 PM
That was SARCASM!! :P

And wrong. Vodafone is an independent network.
Virgin uses Optus' 3G Network. Virgin is part of Optus.

Technically, they're all the same as each other. Don't quote me on this, but...

Telstra has a deal with Three, Three is part of Vodafone, Vodafone has a deal with Optus, Optus has a deal with Telstra, Telstra has a deal with Vodafone... Confused?

They're all as bad as each other!! :(

Back on topic, I think this is balls. Carriers just trying to get away with their crappy networks. If a phone is constantly MAINTAINING a connection, wouldn't that mean MORE load than if it only connects when NEEDED?!
This is balls, I say.

Let me correct that for you.

Optus, Vodafone, Telstra and 3 all operate independent core networks. Core networks are separate to radio networks.

Optus, Vodafone and Telstra all operate their own independent 2G radio networks.

In metropolitan areas there are four 3G radio networks: Vodafone, Optus, 3-Telstra and Telstra NextG

In regional areas there are two 3G radio networks: Vodafone-Optus, and Telstra NextG

Vodafone Australia and Hutchison Telecommunications Australia merged in a 50:50 joint-venture to form a new company Vodafone Hutchison Australia (VHA), this new company owns both the 3 and Vodafone brands in Australia. Neither company "took-over" the other.

Virgin started as a MNVO (Mobile Network Virtual Operator) of Optus, reselling their network. A few years ago, Virgin became a wholly-owned subsidiary of Optus

Even though networks can share a joint radio network, the radio networks are not the bottle - it is the equipment in the base-station which routes the packets and calls, and the bandwidth of the link from the base-station back to the network core.

Mychael
22nd February 2010, 08:35 PM
Oh how wrong I was. My sisters Telstra phone has full 3G reception everywhere in the house. My optus iPhone? Has to be sitting in the edge of the upstairs balcony to get 1 bar of flaky 3G reception.:).



Your sisters phone is also an iphone? We've been with optus for yrs and never suffered reception issues till we swapped to iphones.

Having been with both Telstra and Optus at different times my observation for what it's worth is that in practice there is not a lot of difference, both have their share of poor signal areas just in different places.

asphotos
22nd February 2010, 08:39 PM
how about we use some science.

the feature described above is called

FAST DORMANCY

its a RIM technology, and is used in a number of smart phones.

and for a brief on the technology (3gpp working group presentation) (some warnings about its impact)

from the working group document store

http://www.3gpp.org/ftp/tsg_ran/TSG_RAN/TSGR_45/Documents/RP-090941.zip

TSG-RAN meeting #45 RP-090941
Seville, Spain, September 15 – 18, 2009

Agenda item: 7.3.4.2 – Other Closed Work Items and Small Technical Enhancements and Improvements [TEI8]
Source: Nokia Siemens Networks, Deutsche Telekom, Alcatel-Lucent
Title: System impact of poor proprietary Fast Dormancy implementations
Document for: Discussion & decision
1 Introduction
Always on applications sending frequent small keep-alive messages have been causing excessive terminal power consumption in networks deploying long Cell_DCH and Cell_FACH inactivity timers with no PCH states. Every time a UE wants to send a keep-alive message it ends in Cell_DCH, sends a keep-alive message, waits for the Cell_DCH timer to expire, is moved to Cell_FACH, waits for the Cell_FACH timer to expire and is finally moved to idle. If the keep-alive interval is relatively frequent, say once every 60 seconds, and the inactivity timers are in the order of ten or more seconds, the adverse impact to UE battery life is obvious.
The scenario outlined above has led to different UE manufacturers implementing different kinds of proprietary work-arounds to alleviate the problem. A solution used by some UE manufacturers is identifying the keep-alive messages to the radio layer and after the message is transmitted, sending RRC Signalling Connection Release Indication, which is actually meant to be used e.g. when the UE is powered off or an error at NAS level has occurred. Upon reception of the RRC Signalling Connection Release Indication the RNC is forced to move the UE to idle. Hence problem solved – for the UE.
2 Discussion
There is a drawback in some of the proprietary fast dormancy implementations, which focus on the state transition between active and idle, and do not see the whole picture including the work needed to take the UE back from idle to active. Unless a great care is taken in the details, the networks that support UEs with battery-lifetime-friendly inactivity timers and usage of the PCH state also end up paying a major penalty. In essence the UE seizes control of the RRC state machine leaving no options to the network. In terms of signalling load, Idle Mode UEs transitioning from idle to active just to send one keep-alive message will result in a very large signalling overhead between UE and nodeB, and nodeB and RNC, when compared with a UE in PCH state. Thus it would be of significant advantage to the network signalling load to keep those always-on UEs sending relatively frequent keep-alive messages in PCH state rather than in Idle Mode. Unfortunately some of the popular proprietary Fast Dormancy implementations deny the network from having control over the state that the UE is in and cause a major load to the network.
The above mentioned UE behaviour appears harmless at first glance and the benefits to networks that have not been tuned for always-on applications are obvious. However the reality has proven to be more brutal. When network loses control over the RRC state of the UE and a larger population of devices exhibit this behaviour all the time in the system, the network signallingsignificantly increases. It therefore becomes apparent that the WCDMA/HSPA system capacity will quickly become limited, not by the data throughput capacity the radio can sustain over the air, but rather how much signalling the system is able to accommodate. It is really inefficient for a cellular system to be limited by signalling load rather than by raw radio interface resources which depend more on spectrum availability than system technical design.
3GPP standardized a Fast Dormancy feature in Release 8 that considers system aspects and allows the network to retain control over the RRC state of the UE. The UE is supposed to include a cause value “UE Requested PS Data session end” in the RRC Signalling Connection Release Indication. This can be used by the network to detect that the UE has no more data to send and hence can move it to e.g. Cell_PCH state, instead of idle. It should also be clear in the standards that the RRC Signalling Connection Release Indication is not to be used for fast dormancy-like behaviour without using this specific cause value. In addition the significantly lower latency of resuming user activity from PCH state compared to Idle mode, should not be overlooked when considering the impact to user experience.
It is important to understand that the Release 8 standard based Fast Dormancy can be implemented in pre-Release 8 devices and networks. Therefore if a UE wants to implement a battery-saving feature, it should be clear that the standard based solution is the one to implement.
3 Conclusions
The UE manufacturers are encouraged to consider very carefully before implementing proprietary Fast Dormancy implementations bearing in mind that even if there are networks that do not currently have always-on application friendly configurations, when considering UE battery life. The expectation is that demand will eventually correct that. In the mean time, and for the foreseeable future, devices that seize control of the RRC state machine from the network incur a huge penalty on all the networks they access, regardless of the actual network configuration. So everyone gets hit.
For the benefit of UMTS it is important that UE implementations never release a functional signalling connection for power saving reasons. The intention of the specifications is clear, the RRC Signalling Connection Release Indication always needs to include the cause value “UE Requested PS Data session end” when used for Fast Dormancy purposes.
Accompanying this discussion paper is a CR to TS25.331 [1]. Even if approved, the CR will not on its own prevent harmful ways of using the RRC Signalling Connection Release Indication without setting a cause value, but simply helps to get the standardized Fast Dormancy feature of Rel-8 correctly specified. So the main intention of this document is to avoid that harmful proprietary Fast Dormancy implementations deployed in the field not become commonplace. Finding the right balance between signalling load and battery life will enable efficient use of network resources and have the cell capacity restricted by traffic load rather than signalling load. This will also avoid the operators having to invest a large cost in their network capacities due to a population of devices that are using always-on applications.
References
[1] RP-090942 25.331CR Clarification on Enhanced SCRI approach for fast dormancy

http://www.3gpp.org/ftp/tsg_ran/TSG_RAN/TSGR_45/Documents/RP-090942.zip if you want the juicy channel state change stuff..

now, what do people think?

Shaun R
22nd February 2010, 08:44 PM
Let me correct that for you.

Optus, Vodafone, Telstra and 3 all operate independent core networks. Core networks are separate to radio networks.

Optus, Vodafone and Telstra all operate their own independent 2G radio networks.

In metropolitan areas there are four 3G radio networks: Vodafone, Optus, 3-Telstra and Telstra NextG

In regional areas there are two 3G radio networks: Vodafone-Optus, and Telstra NextG

Vodafone Australia and Hutchison Telecommunications Australia merged in a 50:50 joint-venture to form a new company Vodafone Hutchison Australia (VHA), this new company owns both the 3 and Vodafone brands in Australia. Neither company "took-over" the other.

Virgin started as a MNVO (Mobile Network Virtual Operator) of Optus, reselling their network. A few years ago, Virgin became a wholly-owned subsidiary of Optus

Even though networks can share a joint radio network, the radio networks are not the bottle - it is the equipment in the base-station which routes the packets and calls, and the bandwidth of the link from the base-station back to the network core.

I know all that, but they all have deal & inter-linking with each other. Thanks for clarifying though.

Shaun R
22nd February 2010, 08:46 PM
how about we use some science.

the feature described above is called

FAST DORMANCY

its a RIM technology, and is used in a number of smart phones.

and for a brief on the technology (3gpp working group presentation) (some warnings about its impact)

from the working group document store

http://www.3gpp.org/ftp/tsg_ran/TSG_RAN/TSGR_45/Documents/RP-090941.zip

*SNIP*

http://www.3gpp.org/ftp/tsg_ran/TSG_RAN/TSGR_45/Documents/RP-090942.zip if you want the juicy channel state change stuff..

now, what do people think?

Could you uhhh... summarise that in 2 sentences, please?? No ay am I reading that!

Exocet
22nd February 2010, 08:49 PM
I know all that, but they all have deal & inter-linking with each other. Thanks for clarifying though.
Aside from 3's Roaming agreements with Vodafone and Telstra, I'm unaware of any such "inter-linking" other than the sharing of radio networks I discussed earlier.

asphotos
22nd February 2010, 08:54 PM
Could you uhhh... summarise that in 2 sentences, please?? No ay am I reading that!

and thats the problem............ you don't understand the issues, and refuse to investigate the problem, but blame your balls?

sageco
22nd February 2010, 09:29 PM
and thats the problem............ you don't understand the issues, and refuse to investigate the problem, but blame your balls?

Okay, thats much too harsh a call man, the document and links may be all well and detailed; but there is no way an outsider can possibly understand it. It has numerous acronyms and makes use of technical jargon to such an extent that expecting anyone other than a telecommunications engineer (in general alright-obviously enthusiasts and people involved will know what it means) to understand it is ludicrous.

Still i read it and as far as I can grasp, they are complaining that the way cellphone manufacturers implement fast dormancy conflicts with the way in which the towers determine the device state.

In an ideal world (for the network) the device state would be determined by the towers; however, the manufacturers means of altering the device state causes the towers to lose control of the connections resulting in a wasted lines and a significantly higher signalling rate.

waynie
22nd February 2010, 09:43 PM
Regardless of whether Apple is at fault or the carrier, it's clear that some carriers are almost completely unaffected by these issues, which demonstrates a vulnerability in certain carriers networks. It seems that Optus are the only Australian carrier affected - who is the vendor of their network infrastructure? If an iPhone can cause network disruption through normal operation, what's to stop a denial of service attack using the same methods? Surely it comes down to a question of infrastructure.

Al Aero
22nd February 2010, 09:46 PM
Regardless of whether Apple is at fault or the carrier, it's clear that some carriers are almost completely unaffected by these issues, which demonstrates a vulnerability in certain carriers networks. It seems that Optus are the only Australian carrier affected - who is the vendor of their network infrastructure? If an iPhone can cause network disruption through normal operation, what's to stop a denial of service attack using the same methods? Surely it comes down to a question of infrastructure.

And also the fact that they oversold their network.

ipwn
22nd February 2010, 10:15 PM
Regardless of whether Apple is at fault or the carrier, it's clear that some carriers are almost completely unaffected by these issues, which demonstrates a vulnerability in certain carriers networks. It seems that Optus are the only Australian carrier affected - who is the vendor of their network infrastructure? If an iPhone can cause network disruption through normal operation, what's to stop a denial of service attack using the same methods? Surely it comes down to a question of infrastructure.


Agreed, Its hard to believe, almost 2 years later with all the technologies and interwebs and billions of dollars, they still cant fix the problem. who's head is in the sand??? Apple or the Telcos?

chotus
22nd February 2010, 11:20 PM
A 2 sentence summary of the paper might be:
RIM et al are taking advantage of the grey areas of the standards by introducing proprietary features that are efficient for the individual device, but inefficient at the system-wide level. Due to these features the network can lose control of the device's state, and assumptions made during the design of the network will no longer hold.

amazing what a little reading comprehension will bring you...

6andy6
23rd February 2010, 07:36 AM
Telstra's network blows this theory to shit..

I was on Optus for 18 months and it was nothing more that a complete disaster, no data, dropped calls, delayed messages etc etc.

Been on Telstra since December and have not had one issue.

Well if your location is Stockholm (see right upper), then I am not surprised you had problems with Optus towers - hell of a long way to get a signal!!!!!!!!

Lol :tongue:

The_Hawk
23rd February 2010, 07:58 AM
An interesting read, but it does seem to be slanted towards iPhones (and other smartphones) and their data connections rather than the voice traffic. I can understand how a system that doesn't drop it's port when finished can quickly use up all the "free" ports, but if this is such a huge problem you would have to assume that Telstra and the like would also hit that wall very quickly since individual devices can be allocated many ports if it does something different every 15 seconds and drops the connection in between.

But back to what I wanted to say, and to preference it all, yes on the Optus network.

10 years ago when I first started catching the train to work I could pick up the phone and talk to someone at pretty much any point on the trip (there was always one dead spot)... Now half the trip the iPhone is saying No Service or Searching.

Sure it was always on a Nokia and it was back before 3G, but even so, I was sitting next to a mate on the train the other day with his run of the mill Nokia and guess what?! No service for most of the way home. So we aren't even talking about the inability to make a call even though it's saying you have coverage, just flat out no coverage.

Now I thought the old 2G service was still there as a fallback to the 3G system, so what's the issue? I wouldn't have thought that over subscription would cause these issues, more so a degrading network? Or maybe new building getting in the way of old towers (and no new ones?).

Who knows what the "real" centre of the issue is, but we all know there is an issue.

Al Aero
23rd February 2010, 08:01 AM
Well if your location is Stockholm (see right upper), then I am not surprised you had problems with Optus towers - hell of a long way to get a signal!!!!!!!!

Lol :tongue:

I had better Optus range in Stockholm last year than i have had in Melbourne in the past 18months :p

timmytomtam
23rd February 2010, 05:43 PM
Just moved house and am waiting to get ADSL connected.

In theory, using PDAnet on the iphone would be fine for a few weeks, until we get broadband.

Oh how wrong I was. My sisters Telstra phone has full 3G reception everywhere in the house. My optus iPhone? Has to be sitting in the edge of the upstairs balcony to get 1 bar of flaky 3G reception.

Can't wait till August when I an ditch optus and move to Telstra :)

A quick update on this.

I purchased a Telstra pre-paid broadband dongle today, to keep me satisfied until Internode come round to connect us on the 17th of March.

I'm amazed at the coverage. Everywhere in the house, I have full NextG coverage.

To the question of wether it's an iPhone issue, for me it's just optus - I tested by taking the SIM out of the dongle and putting it in the iPhone and it's obvious that Telstra just have a much better network

iCameron
23rd February 2010, 09:00 PM
Telstra's network blows this theory to shit.

Telstra's network is an example of infrastructure built from the ground up to support 3G data, and the constant connection/reconnection of 3G. Optus however is oversold, and is built to support a 3G voice network, with data tacked on.

marc
23rd February 2010, 09:06 PM
Your sisters phone is also an iphone? We've been with optus for yrs and never suffered reception issues till we swapped to iphones.

Because your old phones weren't 3G?

Mychael
24th February 2010, 01:35 PM
Because your old phones weren't 3G?

Sorry not sure if you meant that as a question or statement.My old phones Sim card was 3G and was put into the iphone. It was all working but signal reception was patchy, I contacted iphone customer service and they recommended fitting a brand new Sim card.
I had this done and got some improvement but not to the reception ability of my previous phone.

When I speak of reception ability I mean in the context of getting good signal to make voice calls. I seem to be finding quite a few places where the iphone is marginal and my previous phone was fine.