25-06-2011 10:48 PM
I realise the engineering teams have been drive testing for a number of weeks. A couple of weeks ago you stated the engineers think they have a couple of reasons why people are losing 3G connection or more importantly connected to the 3G network but lose IP or ‘internet connection’ and or intermittently very slow throughput speeds. The engineering teams have still not found out what is causing the problems or maybe you don’t want tell us?
I have been patiently (and bitting my tongue) observing the Vodafone responses to customers however behind the scenes with a small very advanced virtual team of close friends (in their own time) helped me carry out a number of tests on all three networks. The objective was to compare mobile broadband performance (not sure why Vodafone test teams are taking so long, maybe it’s the same people rolling out the network so little dedicated time and or are not full time on the testing or the analysis of results are being done overseas instead?? ). One interesting observation is the below was much worse on Optus and much minimised (or non-existent) on Telstra.
Background before findings
3G mobile broadband devices (i.e. dongle, 3G WiFi modem) can be in one of the following states in terms of Radio Resource Control (RRC) connection -- strictly speaking, the Idle state is not a Connected state:
High (also described as DCH or Cell DCH or HSPA). In this state, the device is communicating with the network in a dedicated fashion, which requires the most network resources and the highest device power (compared with the other states).
Low (also described as FACH or Cell FACH). In this state, the device communication with the network requires some network resources (also supporting data throughput at low rates) and some device power, but has a lesser burden than the High state.
Standby (Also described as URA, URA PCH or Cell PCH). In this state, the network is aware of the device, but there is no requirement for network resources to be consumed (please note, Cell PCH is similar but very different from URA PCH – I will talk about this later).
Idle. In this state, the network is aware of the device, but the device is in sleep mode, not really communicating with the network and therefore requiring the least amount of network resources and consuming the least power (compared with the other states).
When a 3G mobile broadband dongle or pocket WiFi or embedded device was not transmitting any data, for example when reading a web page or reading an email but not downloading, the network appears to go into CELL_PCH state and not URA_PCH state (maybe Vodafone does not have URA_PCH?). Then when keep alive messages need to be sent, the state is transitioned to CELL_FACH and once done its sent back to CELL_PCH. Now the transitioning back from CELL_FACH (or CELL_DCH) to CELL_PCH can take quite some time.
I did some measurements to get an idea of the time required when switching between different HSPA air interface states. While data is transferred, the mobile broadband device is usually in Cell-DCH state on a High Speed Shared Channel. When only little or no data is transferred, the connection would transfer to the Cell FACH Channel, which is slow but has little overhead for both the network and the mobile modem/dongle in terms of control measurements and power adjustments. If no data is transmitted for a longer duration (e.g. one minute+) it appears in a number of instances the connection was put into Idle state. While the IP address is retained, the physical connection between the mobile network in a number of cases was lost. In a number of other cases the 3G connection between the mobile broadband modem and the 3G base station showed a connection however the ‘Internet Access’ was lost (you would better observe this with an embedded device where ‘Windows Network Sharing Center’ within Windows 7 would show ‘3G connected’ but internet in the pop up window within the tool bar ‘Wireless Connection’ would state ‘No Internet Access’??
The round trip time to the first hop in the network of a ping is around 80 to 150 milliseconds while the mobile broadband device is using a high speed shared channel (which very good and almost on par with Telstra). While on the slower Cell FACH access channel, round trip time would increases up to 300 milliseconds. Returning to the high speed shared Cell DCH channel in some instances took around 500 to 2500 milliseconds!!
The effect of this state switching is that web browsing feels slow over HSPA. When testing the web browsing, I measured the effect of state transition from FACH to DCH on web page load times and depending on network load etc it can have quite an impact. Tests showed an average page load time to the Vodafone home page on a number of occasions greater than 6 seconds. The implication when this happened is the connection sat in FACH for about 5 seconds and struggled to send / receive any data (this was observed in number of suburbs).
Where it is most noticeable is when you load a page, pause a while to read then try to follow a link. If the connection has dropped back to idle / PCH or FACH, you would then perceive a slow connection because of the amount of time it appears for something to start happening. As stated above if you were in Cell-DCH state the data throughput speeds were very good, with high speeds ranging from ~3Mbps to ~7Mbps dependant on the base station being a 7.2Mbps or 14.4Mbps rated.
So in summary, I rarely noticed any transitions to Cell_PCH state, but after a period of inactivity the mode on a number of occasions changed directly to Idle state (not sure if this is suppose to happen on Vodafone or they way we were testing??). This was rather surprising, since standard 3GPP literature points to a transition to Cell_PCH or URA_PCH instead of Idle?? There is a big downside to using the Idle state: When you restart activity it takes around 2 to 4 seconds before data can be exchanged again, something that would be quite noticeable (a solution to this is not putting the mobile device in Idle state but rather in the so called URA-PCH state, more on this in my suggested solutions below).
The standards state the URA-PCH state is quite similar to Cell-FACH with the major difference being that the downlink does not have to be observed continuously. When the mobile broadband device wants to transmit data again, it can immediately go to Cell-FACH state as there is no need to establish a signalling connection and perform an authentication procedure etc. This significantly reduces the time it takes before the first packet can be sent. For the Vodafone network, the advantage is that much less signalling would be required to return the mobile broadband device to a full Cell-DCH high speed state compared to doing the same from Idle.
I was reading through the UMTS 3GPP standards and one of the most talked-about HSPA features is Enhanced Cell FACH (a 3GPP Release 7 DL feature and Release 8 UL feature). The mobile broadband device transition time from Idle will be greatly reduced. In other words, with Enhanced Cell FACH, mobile broadband devices will not require a dedicated channel for receiving / sending small amounts of data, including emails or reading a web page.
Another option would be for Vodafone to use the URA-PCH state instead of Idle state as it appears Telstra does, or should go straight to 3GPP Release 8.10.0. With 3GPP Release 8.10.0, instead of just releasing the signalling connection when it needs to, the mobile broadband device has to wait for the expiration of a network configured timer. Once the timer expires, the mobile broadband device can send a signalling connection release indication message with a new parameter that indicates "UE requested PS data session end". At this point the network can then decide to do nothing, to release the mobile broadband device to Idle or to put the connection into Cell-/URA-PCH state.
Customers with Mobile broadband devices would benefit from this as instead of going to Idle, the network can also put them to URA-PCH state and hence, the latency when returning to a fully active state is much shorter than from the Idle state.
To speed up the transitioning to CELL_PCH state in Release-8 when the mobile broadband device sends SCRI message, it’s supposed to include the cause value as "UE Requested PS Data session end". Once the network receives this cause it should immediately move the UE to CELL_PCH state. The good thing is that even pre-Release 8xx mobile broadband devices can still implement this feature and it would still work.
Tim, let me be clear, I understand you are not going to tell me or others how the network is configured and nor do I expect any answers to the above! It is a wireless network thus open to objective comparison testing and since we don’t live in a communist country we have freedom of expression and opinions. Let’s face it, if everything was of the expected standard of a mature country and mature network then this blog would not be needed!
Like I have stated before in another blog in-building 3G tests have shown where there is 850MHz turned on, the in-building coverage and throughput speeds (like for like devices ie 7.2Mbps only devices) I found to to be similar to Telstra (Optus does not have a 3G 850MHz network thus poor in-building in a number of locations tested) and I commend Vodafone for this in the space of 6 months!
You now have to escalate to management and have a different in country ‘dedicated 3G engineering crack team’ focussed on quality and to very quickly test and the mobile broadband IP issues. The turnaround times on this site for mobile broadband loss of internet (not relating to coverage) is appalling.
I mentioned in another blog, passive 3G mobile broadband devices (constantly pinging and running remote commands and used by another network) can be deployed in your network identifying 3G and quality issues well before Vodafone customers finds out (not sure if my advice was even escalated or perused by D level management). I doubt you would find a more passionate group of people on this blog site (too many advisors to mention) giving advice to Vodafone. You are building a great 3G/UMTS 850MHz network to complement the 2100MHz frequency, which WILL surpass Optus, just try and do things a smarter way and not the old fashioned way (i.e. drive testing) if people inside won’t change, change them.
I agree each customer needs to fill in the fault template to determine what their fault is related to but 300 non-technical customer service staff taking calls relating to non-coverage, intermittent and constant loss of internet complaints and still relying on old fashioned drive tests (still need drive test until you start installing passive devices) to identify the above and taking weeks for answers is not what I and your loyal mobile broadband customers expect of the biggest mobile network, in the world, Vodafone!
27-06-2011 03:16 PM
Thanks for your detailed post. I've made sure it's passed to the engineers involved.
Just to clarify, I'm not able to provide updates on every step that changes during the process, our network team are busy running the network, and identifying suitable resolutions for the specific issues being discussed here on the Community, and updates are communicated through those engineers to our staff here and to myself.
Much of the technical data isn't shared as the overall relevance for why something isn't performing as well as we expect isnt relevant to a lot of our customers.
The simple version is: Some of our customers services arent performing as expected, identify and correct that as quickly as possible.
That being said, there isn't a point where more information isnt useful, and while I am aware of a lot of the technical data being used by our engineers, the information you've shared can only assist.
Thank you for taking the time to both locate and to post the information, we appreciate your efforts, and I'll continue to share information via this thread here.