- This topic has 21 replies, 1 voice, and was last updated 16 years, 5 months ago by
.
-
Posts
-
Hi all!
It is often said that a handover can result in the loss of 20 msec speech. And this is the reason for poor speech in case of excessive handovers. That is why an RF engineer usually try to have as less as possible handovers.
What is the doubt here?..
How many TCH bursts are stolen to complete a HANDOVER.
I think one TCH burst if stolen cannot result in a loss of 20 ms speech. If eight TCH or Normal bursts are stolen then it will leads to a loss of 40 msec speech.
If 4 are stolen then it will lead to 10 ms loss from one SPEECH BLOCK and 10 msec loss from NEXT SPEECH BLOCK, therefore a total of 20ms.
Regards
a ho is completed in about 20ms to 100ms. 20ms is a really fast value, which is measured when the MS is performing a synchronized HO (towards a cell synchronized with previous serving cell)
the loss of tch is not only due to FACCH, but also the time the MS takes to reconnect to the target cell (ho access, TA calculation, etc)
thanks for reply
but
i haven’t found my answer.
Ok
what is the delay for a MS to MS call.
somebody saying it is 80ms.
I was of the opinion that it was 20msec.
Do you agree.
Regards.
read twice.
delay is 20ms to 100ms, depending if the HO is synchronized or not.Delay in MS to MS call is one question.
Time required by a handover to get completed is another question.
you are saying that a handover usually takes 20 to 100 ms. OK AGREED.
But don’t you think somewhere we are getting the things mixing up.
I have two doubts..let me put them together.
1)How many TCH bursts needs to be stolen for completing a handover.
2) What is the delay for a MS to MS call.
For the first you have said indirectly that handover needs 20 ms to 100 ms.
For the second case you said nothing.
Coming back to first one….
Can we go like this..One TS =.577 micro sec
Min time for handover=20ms
Max time for handover=100ms
TDMA frame=4.615 msecfour TDMA frames will be roughly equals to 20 msec. In this case iam taking the 4.615 ms-.577micro sec time as included for 29 msec.
Roughly we can say Handover needs min 4 and max 20 TCH bursts to be stolen to complete the handover?
Also for losing a complete 20 msec speech block one needs to loose all the 456 bits corresponding to the 20 ms speech. But These 456 bits are never sent on one TCH burst rather they needs 8 bursts(8 x 57=456). The right side of TCH contains 57 bits from N+1 speech block.
Hence i think we cannot say any thing with surety like a handover will result in a loss of 20 ms speech( which is common myth).
simply(as per your points) we can say that a handover takes a time of 20ms to 100 ms to gets completed and thus “how much speech” is lost during this period that is a calculation and is variable.For the delay of 80 msec for a MS to MS call a friend of mine has put up a nice theory(seems correct to me)
20 msec of delay comes at the RPE-LTP vocoder.
20 msec of delay comes at Interleaver.
So total delay at MS=40 ms
This call when switched by MSC comes to a transcoder. There the transcoder again performs RPE-LTP vocoding and we gets a delay of 20 msec.
Then at BTS the interleaver again induces a delay of 20 msec.
Thus ignoring transmission delays it comes out to be
20+20+20+20=80 ms
Regards
Pix
please don’t take anything as seriously. From you last post i feel i have irritated you.
sorry
and
regardsMKT.
hi mkt 🙂
no i’m not irritated at all, don’t worry. i’m just answering very directly to your posts because it goes faster. sorry about that, i hope i didn’t upset you.thanks for the explanations, it’s a bit clearer. Well, at least it’s clear that we’re misunderstanding each other, and it’s gonna be hard to settle this via internet…
do we agree that a TCH timeslot is stolen by the FACCH only to transmit the “HO COMMAND” message ? a HO COMMAND is a very short message so it uses only one or two bursts, max (i could check the message length in 3gpp … but no time 🙂 )**** that’s the answer to the question (1) ****
So the FACCH is not the cause of the “blank” in the voice.
a HO duration is counted from the reception of a HO CMD to the first TCH is sent to the target cell : that is where the delay is : the MS has to access the target cell.
When the MS is connected to the target cell, it will start receiving the current speech frame, as being transmitted by the transcoder. I’m not sure why there would be delay in the TC, the TC works in real time. If a speech frame is not transmitted on the radio, then it’s lost forever.
***** regarding question (2) *****
i don’t understand (i’m sorry !! 😀 ). delay for MS-MS call : are you talking about call setup time from MS to MS ?Hi pix
thanks for your reply
Second question.. here it is
Suppose i want to make a phone call to you.
I dial your MS from my MS.
Connection is establish.
we are talking.
What iam speaking is heard by you 80 ms later. This delay is question here.
Voice is first collected for 20 ms and then processed by vocoder. This gives a delay of 20ms
After x y z we get 456 bits corresponding to 20ms speech. These 456 are then divided into 8 blocks(8 x 57=456). These blocks are then sent one by one over “normal” burst. The first 57 bits of normal burst represents the 20 ms speech block no N, and the second 57 bits of normal burst represents the 20 ms speech block no N+1,.
For this we have to wait for the N+1 speech block to arrive. This again induces a delay of 20 ms.Thus interleaving induces a delay of 20 ms.
so total delay at my MS is 40 msec.
My call reaches the MSC.
MSC send it towards you.
At transcoder again we will have 64/16 conversion ie RPE-LTP which gives 20ms delay.
At BTS the interleaving takes place. and this again induces the delay of 20 msec.
hence the total delay will be
20+20+20+20=80 msec.
This theory doesn’t includes transmission delays.
Is it ok?
Also i don’t know reverse of vocoding and de-interleaving will induce the equal delays or not?
If yes..then
MS to MSC total delay we have
20+20 at MS(vocoding+interleaving)
20 at BTS —deinterleaving
20 at transcoder for reverse of vocoding
This gives 80 ms in one direction.
considering equal delay in other end direction
we have 80+80 =160 ms delay for a MS TO MS CALL.
Just for comments from every body
Regrads
mkt,
ok, i understand which delay you’re talking about now 🙂 well, your explanations make sense, but 80ms of delay is so short that if you hear a signal and the same signal delayed by 80ms, they sound so close to each other that it’s almost instantaneous.
whereas when you call someone which is in front of you (that’s stupid, but just for the experiment), the difference between the natural sound and the MS sound is more like 500ms to 1000ms.
please if you know.give me the document about detail of SACCH multiframe
sorry pix I want ask you about that
mohsin,
SACCH multiframe… give a look at 3GPP 45.00x, the explanations are there.
please explain I am sorry I do not understand.please answre me about the SACCH multiframe.
thank you
Mohsin
1/7/2008please explain I am sorry I do not understand.please answre me about the SACCH multiframe.
thank you
Mohsin
1/7/2008
- The forum ‘Telecom Design’ is closed to new topics and replies.
