Receiver sensitivity- Packet Detection v/s Packet capture

Hello All,
There is a notion of "detecting the signal" and "capturing the
signal" based on the received signal strength. The packet reception
success depends upon the noise on the channel too (SNR). For simplicity
of this discussion, let us ignore noise for the time being.

According to the best of my knowledge, the only physical entitiy
present(which impacts the discussion) in the data sheet of a wireless
receiver is the "receiver sensitivity". If the received signal level is
higher than this sensitivity (noise ~= 0), then the receiver is able to
successfully receive the packet.

However, in certain research work and simulators, "capture threshold"
and "reception threshold" variables are also introduced. Here is the
main point of this topic. Are these abstract entities? In a simple
scenario where a receiver is placed at some distance from transmitter.
Is there a possibility of sensing the packet but not being able to
decode it (assume noise ~= 0 again) ? If not, then the whole idea of
capture threshold is meaningless. If yes, then why doesnt the wireless
cards have it in their specs?

My thinking is that a signal can be captured properly if the
demodulation phase succeeds. There is always something on the channel.
If the demodulator can figure out the envelope wave (depending on the
modulation type), it should be able to demodulate the wave and capture
the digital content from the analog signal. Is there a possibility that
we detect the envelope wave but still not able to demodulate? (Again
assuming noise ~= 0). If this is so, then there can be carrier sensing
but incorrect reception.
Since the whole envelope wave is sent with the same power, is there any
reason of not decoding it after detecting it.

Wlog, I guess we can assume that this effect can be stated even when
any noise present on ths channel. There is still capture/detect effect,
but now we have to consider SNR ratio instead of just signal strength.

Any pointers to this topic is greatly appreciated. Sorry if the
question is very trivial to DSP/communication people!

Thanks and Regards,
Vinay

On 13 Oct 2005 11:52:05 -0700, "Vinay" <(E-Mail Removed)> wrote:

> There is a notion of "detecting the signal" and "capturing the
>signal" based on the received signal strength.

Not really. In Wi-Fi, receiver sensitivity is measured at a specific
BER (bit error rate) which implies a specific signal to noise ratio
ratio Eb/No.
http://www.sss-mag.com/ebn0.html
Generally, the reference BER is either 10^5 or 10^6.

Signal "capture" is an FM effect that decrees that FM signals over a
specific margin (usually 1-2dB) will completely "capture" or cause the
lower level signal to disappear. This works with 802.11 1-2 mbits/sec
which is pure FM, but not with higher speeds, which include an AM
component.

>The packet reception
>success depends upon the noise on the channel too (SNR). For simplicity
>of this discussion, let us ignore noise for the time being.
>
>According to the best of my knowledge, the only physical entitiy
>present(which impacts the discussion) in the data sheet of a wireless
>receiver is the "receiver sensitivity". If the received signal level is
>higher than this sensitivity (noise ~= 0), then the receiver is able to
>successfully receive the packet.

Wrong. The SNR must be above a specific margin or the demodulator
will belch garbage (errors). The measurement reference SNR is NOT a
good place to operate. At 1 error every 10^5 bits with 512 byte
average packets, that's one error every 250 packets, which sucks. With
a 10^5 BER, it even worse with one error every 25 packets. + That's
what "fade margin" is all about. A minimum fade margin of 20dB is
typical, which yields a much better BER for adequate communications.

> However, in certain research work and simulators, "capture threshold"
>and "reception threshold" variables are also introduced. Here is the
>main point of this topic. Are these abstract entities?

Neither term has any real meaning to me. I'll guess that reception
threshold is acutally minimum detectable signal and capture threshold
deals only with FM modulated signals. Neither of these have any EIA
or ISO specs for their measurment.

>In a simple
>scenario where a receiver is placed at some distance from transmitter.
>Is there a possibility of sensing the packet but not being able to
>decode it (assume noise ~= 0 again) ? If not, then the whole idea of
>capture threshold is meaningless. If yes, then why doesnt the wireless
>cards have it in their specs?

You were right the first time. Receiver threshold is meaningless as
it really a measure of how much ambient noise is in the local
environment and how much signal it takes to overcome this noise. Not
a very useful or reproduceable measurement. Instead of 0dB S/N (which
is where the signal equals the noise) a BER reference makes more sense
and is reproduceable.

> My thinking is that a signal can be captured properly if the
>demodulation phase succeeds.

Captured properly? That implies an interfering signal.

>There is always something on the channel.

What does that mean? Are you making measurements with an antenna
connected? If so, you won't get consistent results. The receiver
sensitivity should be measured in a pollution free environment with a
reproduceable test setup. Ambient RF pollution is not reproduceable.

>If the demodulator can figure out the envelope wave (depending on the
>modulation type), it should be able to demodulate the wave and capture
>the digital content from the analog signal.

Yeah, that's what demodulators do. With direct sequence spread
spectrum, it's a bit different as there is no analog signal. Most of
the common chipsets used today are all digital and use a DSP to
provide the necessary digital demodulation without first converting to
analog.

>Is there a possibility that
>we detect the envelope wave but still not able to demodulate?

Sorry. You lost me.

>(Again
>assuming noise ~= 0).

Again, nobody trys to demodulate a high speed digital signal with 0dB
SNR. The higher data rates and more exotic modulation schemes require
a much higher SNR. For example, at 11Mbits/sec with CCK modulation,
the minimum SNR is about 11dB for a 10^6 SNR. Look at the curve at:
http://www.sss-mag.com/ebn0.html
There are modulation schemes that will work with 0dB SNR, but the
thruput isn't anything near as fast as Wi-Fi.

>If this is so, then there can be carrier sensing
>but incorrect reception.
>Since the whole envelope wave is sent with the same power, is there any
>reason of not decoding it after detecting it.

You lost me again.

>Wlog, I guess we can assume that this effect can be stated even when
>any noise present on ths channel. There is still capture/detect effect,
>but now we have to consider SNR ratio instead of just signal strength.

You most certainly should consider SNR. If insufficient, you get
noise on your demodulation output instead of data.

> Any pointers to this topic is greatly appreciated. Sorry if the
>question is very trivial to DSP/communication people!

You're probably asking in the wrong newsgroups. I can't tell what
you're working on or what you're trying to accomplish to advise as to
the proper newsgroup.

--
Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
831.336.2558 voice
http://www.LearnByDestroying.com AE6KS
http://802.11junk.com Skype: JeffLiebermann
(E-Mail Removed) (E-Mail Removed)