| Author |
Message |
fl
Guest
|
 Posted:
Sat Dec 24, 2005 1:16 am Post subject:
Question about Pr vs. BER |
|
|
Hi,
Some papers use a plot to designate the percentage time of time which
is above a BER data. Although it seems clear to me, I have no idea to
do my simulation to get it. I have browsed the book:Simulation of
communication systems, written by Michel C. Jeruchim etc. and not find
the related topic. Now I am interested in WLAN system, a packet based
transmission mode. I want to evaluate the system performance besides
the normal BER figure.
Could you shed some lights on this?
Thank you in advance |
|
| Back to top |
|
 |
Mike Yarwood
Guest
|
| Posted:
Sat Dec 24, 2005 8:21 am Post subject:
Re: Question about Pr vs. BER |
|
|
"fl" <rxjwg98@gmail.com> wrote in message
news:1135375005.625304.218600@g43g2000cwa.googlegroups.com...
| Quote: | Hi,
Some papers use a plot to designate the percentage time of time which
is above a BER data. Although it seems clear to me, I have no idea to
do my simulation to get it. I have browsed the book:Simulation of
communication systems, written by Michel C. Jeruchim etc. and not find
the related topic. Now I am interested in WLAN system, a packet based
transmission mode. I want to evaluate the system performance besides
the normal BER figure.
Could you shed some lights on this?
I've not seen any of these papers - could you post some references? Thanks. |
I'd be interested in anything you find.
Best of Luck - Mike |
|
| Back to top |
|
|
fl
Guest
|
| Posted:
Sat Dec 24, 2005 5:16 pm Post subject:
Re: Question about Pr vs. BER |
|
|
Hi,
At hand, this is one:
IEEE trans. on Communications, Vol. COM-31, No.9, 1983,
"Frequency-selective fading effects in digital mobile radio with
diversity combining",
Page 1085-1094, Bernard Glance and Larry J. Greenstein.
Fig. 2-7
Thank you. |
|
| Back to top |
|
|
Eric Jacobsen
Guest
|
| Posted:
Sat Dec 24, 2005 11:44 pm Post subject:
Re: Question about Pr vs. BER |
|
|
On 23 Dec 2005 13:56:45 -0800, "fl" <rxjwg98@gmail.com> wrote:
| Quote: | Hi,
Some papers use a plot to designate the percentage time of time which
is above a BER data. Although it seems clear to me, I have no idea to
do my simulation to get it. I have browsed the book:Simulation of
communication systems, written by Michel C. Jeruchim etc. and not find
the related topic. Now I am interested in WLAN system, a packet based
transmission mode. I want to evaluate the system performance besides
the normal BER figure.
Could you shed some lights on this?
Thank you in advance
|
I think I know what you're describing.
For bursty, packet-oriented systems like WLAN the BER metric is not
very useful. A packet has to be received error-free or it won't get
ACKed, so Packet Error Rate is generally used instead.
WLAN channels are frequency-selective fading channels, and in most of
the stochastic channel models the performance degradation is dominated
by the worst channel instances that may be part of, say, 1000
instances generated using the stochastic fading models. In other
words, you generate a thousand or so random instances according to the
characteristics of the particular model, and then cycle through or
randomly select them to test your system against. If you just plot
PER vs SNR for all of the instances, the performance will often be
much worse than if you plot PER vs SNR for the 90% best channels.
Often the 90% best channels will provide a better representation of
how the system will really behave if you're trying to evaluate
something like acquisition or FEC. Basically, you're just chopping
the bottom off of the distribution since otherwise that small part of
the distribution will dominate the system performance. If you keep
the bad channels in you can mask effects of the real system to the
point that you'll never know whether you really have system problems
or not.
It's kind of like importance sampling, I guess.
Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org |
|
| Back to top |
|
|
James (Sungjin) Kim
Guest
|
| Posted:
Sun Dec 25, 2005 1:16 am Post subject:
Re: Question about Pr vs. BER |
|
|
Eric Jacobsen wrote:
| Quote: | For bursty, packet-oriented systems like WLAN the BER metric is not
very useful.
|
So, in some paper outage probability in terms of capacity is used. You
can test your system which how frequently dose not achieve the threshold
rate. That is, check
P_out = Pr( capacity( h) < R_threshold)
where h is channel metric, capacity( h) is achievable rate with respect
to the give channel matrix h, and R_threshold is the given threshold
rate. As R_threshold goes higher, P_out becomes larger and larger. If
your system operates with multi-level modulation and high performance
channel coding, you can use the following capacity function
capacity( h) = log_2( 1 + |h|^2), if log_2( 1 + |h|^2) < R_max
or R_max, otherwise
where R_max is the rate provided by the maximum supportable modulation.
If you are interested in this performance measurement, I can introduce
to you some related papers. |
|
| Back to top |
|
|
|
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in