- This topic has 26 replies, 1 voice, and was last updated 8 years, 12 months ago by pix.
29th February 2008 at 15:36 #51264AlexGuest
What action should be taken to improve SQI?
What action should be taken to reduce FER?1st March 2008 at 12:39 #51265PixGuest
Less handovers, less interference (co channels and adjacent channels).
Use of frequency hopping also, to reduce the fast fading effects…1st March 2008 at 16:22 #51266AlexGuest
Can we say the action plans to improve each of them are the same or there are different actions to improve them?
Supose we improve the SQI by some actions,I mean :”Will the FER improve by that?” and vice versa.2nd March 2008 at 07:21 #51267BilalGuest
Better codec i.e. EFR and AMRFR?2nd March 2008 at 10:06 #51268PixGuest
A better FER will certainly lead to a better SQI : the FER is part of the SQI computation. FER is mostly depending on the radio quality and the codec’s protection to interference.
But a better SQI can be achieved also with a different approach, such as less handovers, or better codecs, as suggested by Bilal.
A paradox : “Better” codec offers less protection than a “resistant” codec… The former will improve voice quality under good radio conditions, the latter will improve voice quality in a poor C/I situation.
Pix2nd March 2008 at 17:32 #51269akefaGuest
why do you hate HO so much 🙂
Did you mean from saying “less HO leads to better SQI” that you want calls to drop or at least to be carried by the bad quality carriers? Please dont
The subscriber has no idea how many HO is going on.
In TEMS, I usually ignore the bad Rxqual and SQI in the moments of HO’s because it is not the subscribers precieved ones. dont you?
I think the performance of HO’s is more important than the number of them.2nd March 2008 at 17:49 #51270PixGuest
Handovers are obviously needed to keep a satisfying radio link… But I’m talking about “useless” handovers.
Basically, what I’m saying is : reduce the number of power budget handovers, and apply a light ping pong hysteresis. This can be done by using a longer PBGT HO averaging window, and higher HO MARGIN(0,n).
The limit of this, as you’re pointing it out, is when the call drops increase or too many emergency HO are taking place. But don’t forget than most users are motionless, and therefore do not need to do too many handovers.
That answers one thing. Second thing was : HO vs. voice quality
For the users, a handover is heard as a blank of 40ms to 80ms in the voice. It is not perceived consciously, but it degrades the speech quality. In SQI or VQ, a handover gives a very bad mark 🙂 Therefore, if you’re being compared to a competitor then you might have a surprise.. a bad one 🙂 even though call drops are low and C/I is high, the number of ho will be decisive…
Some of our clients were deeply disappointed by VQ’s match results..2nd March 2008 at 18:53 #51271akefaGuest
first of all dont forget: there is no a unique measuer for the sqi. It is vendor dependent.
Recently there is a new speech quality indicator specified by ITU called “benq” for all vendors.2nd March 2008 at 19:51 #51272PixGuest
SQI is provded in TEMS, i didn’t know there were other indicators with the same name…
Do you have details about Benq ?2nd March 2008 at 20:19 #51273AlexGuest
As akefa said according to the specification of GSM, it is not possible to get information above
the real speech quality, as perceived by the subscribers. It is only possible
to estimate the subscriber perceived speech quality.
But in R8 Ericsson system, speech quality supervision was introduced which gives statistics based on
SQI samples reported to the BSC.
Also I should remind you in TEMS, measuring SQI is included by all necessary factors except Speech Codec.I think this measured SQI is good to rely on.
Alex3rd March 2008 at 05:32 #51274BilalGuest
Speech Supervision stats in ericsson BSC are based on FER only, nothing else. Samples are ranked as GOOD, ACCEPTABLE or BAD based on FER thresholds.3rd March 2008 at 16:04 #51275pixGuest
in alcatel, it is the same : frame erasure only. There is the FER and the consecutive frame erasure number (CFE).3rd March 2008 at 19:56 #51276akefaGuest
Pix , SQIs of the TEMS are only the 1st step to measure the subscriber preceived quality. In other words, It is a row data that needs more treatment, for example subtracting the number of HOs from the bad SQI (TEMS considers a bad SQI during HO while the subscriber has at least an acceptable one).3rd March 2008 at 20:26 #51277PixGuest
So actually, not hearing voice for 80ms is acceptable and should be considered with a SQI = 0 ? 🙂4th March 2008 at 21:58 #51278akefaGuest
I am talking about the statistic. for example:
1- good sqi samples=60
2- accpt sqi samples=20
3- bad sqi samples=20
HO numbers= 10
then the real statistic is
good = 60+10