On Feb 14, 2008, at 1:37 PM, Iñigo Urteaga wrote:
Hi all,
I found this previous post from Dec. 19th in tynios-help and I would
like to have some points clarified.
First, I agree with Phil that the received signal strength will be
(assuming waves are in phase) the addition of both desired signal's
power and existing noise power level.
As far as I understand TOSSIM's code, no radio stack is simulated. All
the radio communications are simulated by TossimActiveMessageC, which
relies in CPMModel. CPMModel computes the packet's received power with
it's corresponding noise level using CPM gain/noise model. At this
point is calculated the "strength" value as mentioned by Tal, which is
included in the metadata of the packet. This "strength" field can then
be accessed by TossimPacket interface.
So, having this scenario, I understand the goal of TossimPacket
interface is to give the application a chance to access radiolink
quality information, am I right? Accessing received message's strength
value, it is possible to have a rough idea of the link quality between
the sender and the receiver.
However, I am not completely comfortable with using signal strength as
link quality indicator. Although I agree received signal strength is
the sum of both the original signal and noise, I don't think this
truly reflects link quality. It would be possible to have really noisy
links (with low desired signal power level) that would have as good
strength values as really good links (where desired signal power is
much bigger than noise level). Instead, since signal's power level and
noise level are accessible by the CPMModel component, why not use
Signal To Noise Ratio?
That is why I would like to know if there is any reason that makes
necessary to have the strength value included in the metadata. Would
not be possible to include the SNR value in that field? My point here
is that, since no radio stack is simulated in TOSSIM, it would be more
useful to offer a SNR indicator than the "strength" field.
Of course SNR is a better indicator than S+N. But all most radios let
you do is sample the RF power, which is inherently S+N. The CC2420
provides an indirect measure of SNR through its chip correlation
indicator (commonly but incorrectly called LQI, for link quality
indicator). This measures the soft chip correlation across a couple
of symbols. LQI is pretty good as a very lightweight way to see if a
link has a good SNR, but as you point out, if the link has varying
SNR then you don't get it on bad periods. So it turns out that a
cross-layer approach -- uses layer 1 feedback as well as layer 2
delivery -- generally works better.[1]
Phil
[1] Rodrigo Fonseca, Omprakash Gnawali, Kyle Jamieson, and Philip
Levis. "Four-Bit Wireless Link Estimation." In Proceedings of the
Sixth Workshop on Hot Topics in Networks (HotNets VI), 2007. http://
sing.stanford.edu/pubs/hotnets07-4b.pdf
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
