RE: stds-80220-requirements: latency:ad Hoc meeting yesterday
Dear Masaaki San:
Thanks very much for your comments regarding the requirements for latency
and error rate. I am sorry for the late response because I had a problem in
accessing my email remotely during the meeting in Vancouver.
If I understand your concern correctly, you were probably worrying about the
increase in latency that might occur in a bad channel environment, as caused
by an increase in re-transmissions of the error packets.
If the system is not designed to work in a worse channel environment, there
could be a relatively high error rate and increased latency as the channel
gets worse. At that time, would this system still be able to serve the
mobile user based on the requirements of the applications that the user
wished to run?
The purpose of the requirement is really to specify some acceptable values
of error rate and latency such that the system (PHY and MAC layers) would be
able to provide support on various applications that the mobile users may
use.
In the original proposed text discussed in November's meeting, the use of
the table is mainly to ensure that the future 802.20 standard would be able
to handle different levels of QoS, in terms of packet error rate and
latency, optimized for different classes of applications. The consideration
was from the end user's perspective.
It is true that it may be difficult to determine a fixed latency value for
the 802.20 standard because the network delay is an unknown variable.
Therefore, the values in the table were tentative values based on some
references by ITU, 3GPP etc. In addition, we have updated the proposed text
to a more generic form without including fixed latency values. In the
evaluation criteria, it would then be necessary to determine some specific
values based on assumptions on the typical network delay.
Your examples of system 1 and 2 would be a design tradeoff, which is
depending on the system requirement or performance targets, i.e., the level
of QoS in terms of error rate and latency, variations in channel environment
that the standard is designed to support.
Please let me know if this helps to answer your questions.
Best regards,
Anna.
-----Original Message-----
From: owner-stds-80220-requirements@majordomo.ieee.org
[mailto:owner-stds-80220-requirements@majordomo.ieee.org] On Behalf Of YUZA
Masaaki
Sent: Wednesday, January 14, 2004 1:21 PM
To: Requirements Reflector
Subject: Re: stds-80220-requirements: latency:ad Hoc meeting yesterday
yuza@NEC-i
I mistaked "error rate" for "error late".
Please correct "error late" to "error rate"when you read
my e-mail.
Thank you Wu-san for pointing out my mistake.
********************************
YUZA Masaaki
NEC infrontia Corp.
E-mail:yuzam@pb.jp.nec.com
tel:+81-44-820-4682(personal)
tel:+81-44-820-4545
fax:+81-44-820-4555
********************************
----- Original Message -----
From: "YUZA Masaaki" <yuzam@mwb.biglobe.ne.jp>
To: "Requirements Reflector" <stds-80220-requirements@ieee.org>
Sent: Wednesday, January 14, 2004 10:29 PM
Subject: stds-80220-requirements: latency:ad Hoc meeting yesterday
yuza@NEC-i
Good morning, Anna and every attendees yesterday's Requirement documents ad
Hoc meeting.
I think after the meeting, that we should how to evaluate systems which
proposed.
We discussed and agreed that error late define at MSAP interface at
Singapore.
This means error late is not increase vs propagation loss of air interface.
Which parameter is increased vs propagation loss?
It is latency.
Latency becomes very important prameter of systems for evaluate.
When we design a system, how to behave error late and latency vs propagation
loss(PL).
For exampre,
1.Very good error late characteristics vs propagation loss using powerfull
error correcting code.
2.Fast ARQ!
But each method use redundant bit for decrease information bit.
System 1, using error correcting code, behave not increase mean latency time
vs PL
when it can correct error, and latency is suddenly increase more bad PL
environment.
Other hand.
System2, using fast ARQ, mean latency increase moderate vs PL.
I don't know which latency manner is better?
Where the cross over point that mean latency time system1 over system2?
I think system keeps low mean latency time for wider PL environment.
Then, mean latency is evaluate item of system, not requirement improve
methode.
Best Regards.
Masaaki Yuza
********************************
YUZA Masaaki
NEC infrontia Corp.
E-mail:yuzam@pb.jp.nec.com
tel:+81-44-820-4682(personal)
tel:+81-44-820-4545
fax:+81-44-820-4555
********************************