Hello,
This is my first post in the community. I have been thinking about doing
this question, but after having reviewed some of the theory with digital
signal processing, usrp architecture and some approaches for
positioning/raning, I think I need to ask to the expert community,
mainly to
find a line of work and to know the restrictions for the application we
want
to carry out.
I would like to measure the distance between two radios at short
distances
(we have two UN-210 and two XCVR2450 or 1 WBX and 1 DBSRX2, let’s say to
1
m, 2m…up to for example 20 meters), starting from off-line
experiments
and finishing with real time measures in line of sight. My questions
are:
- What would be, from your experience, the best approach to do this
(transmitted signal and digital processing in the receiver part), what
accuracy should I expect in line of sight?
I am looking mainly for a starting point, the signal structure to
transmit
and the receiver structure. For example, one of the thoughts I have is a
correlation in the receiver of a pseudorandom (spread spectrum)
transmited
signal, this approach could serve to measure time of arrival by
measuring
the delay of the pseudo-noise code signal, when comparing it with an
exact
copy at the receiver. Another approach I have seen is the estimation of
the
Channel impulse response (CIR), that maybe (as far as i know) could
serve
also for estimating distances. I have seen that there is some code to do
a
correlation inside the FPGA of the USRP1.
So far my references are (which maybe can be helpul for those who are
working in the same field):
-
“Advancing Wireless Link Signatures for Location Distinction”. Here,
they
do not carry out ranging/positioning, but a change in the transmitter
position is detected. In my case (in the case ranging is not possible),
to
detect if a distance is crossed would be my objective (e.g. the
transmitter
go closer than 1 meter to the receiver).
-
“RF Ranging for Location Awareness”. In this dissertation they use an
arquitecture similar to the USRP hardware.
As you can see what I am looking for is for some sort of radar/ranging
application and some starting point. Also, my questions are very generic
for
now and I apologize for it, mainly because I am still looking how to
focus
the problem.
Then, any help (references, similar work done) will be very useful.
Kind regards,
Mario
I think it is going to be a challenging project. Note that I only happen
to
know a bit about ranging, but I have no experience at all implementing
this
on GNU Radio/USRP, so I may be off-base at some points.
Your basic idea (correlating against a PN sequence) could work well as a
first step (detecting time of arrival, but it is only one component of
what
would be needed. Time of arrival alone will not give you the distance
without the precise time of transmission.
Your options are Time of Arrival (ToA)/Time Difference of Arrival
(TDoA),
Angle of Arrival (AoA), Received Signal Strength (RSS)-based distance
estimation, or Two-Way ToA/Ranging (TW-ToA). It seems to me that you
essentially have to use TW-ToA, basically measuring the “time of flight”
of
the signal to and from both radios like a ping. You cannot use ToA or
TDoA
because they assume the time of transmission is precisely known, and you
would not have sufficiently synchronized clocks. (Unless you use the
same
clock reference for both USRPs, in which case its not really practical
for
ranging since both USRPs are basically stuck next to each other.) For
AoA
you would need directional antennas, and it does not give range anyway.
For
RSS-based distance estimation, you would need a lot of training data to
correlate RSS readings with distances.
So suppose you need to measure the time difference in both directions,
i.e.
radio A transmits at T1, radio B detects and responds, radio A detects
the
response at T2, and range ~= c*(T2 - T1)/2. This is where I think you
will
have complications… the processing delay introduced by GNU Radio while
detecting and responding will lead to severe inaccuracies, because at
the
speed of light even a fraction of a microsecond results in an error of
dozens of meters. Using in-band signaling and transmit and receive
timestamps could alleviate this problem… However, each USRP should
know
what the processing delay is between receiving a ping and responding to
it.
Maybe it can be kept constant, so that one USRP does not need to inform
the
other one how much processing time was spent? Just a thought.
As for references, I think it may be worthwhile to look at IEEE
802.15.4a.
It defines a precision wireless ranging protocol using TW-ToA, to an
error
of about ~1m. It may be helpful to look at that. Googling for “IEEE
802.15.4a ranging” turned up a few interesting papers. The MERL report
looked pretty good.
Kunal
Mario R. [email protected] writes:
-
What would be, from your experience, the best approach to do this (transmitted
signal and digital processing in the receiver part), what
-
“Advancing Wireless Link Signatures for Location Distinction”. Here, they do
not carry out ranging/positioning, but a change in the
transmitter position is detected. In my case (in the case ranging is not
possible), to detect if a distance is crossed would be my
objective (e.g. the transmitter go closer than 1 meter to the receiver).
-
“RF Ranging for Location Awareness”. In this dissertation they use an
arquitecture similar to the USRP hardware.
As you can see what I am looking for is for some sort of radar/ranging
application and some starting point. Also, my questions are very
generic for now and I apologize for it, mainly because I am still looking how to
focus the problem.
Then, any help (references, similar work done) will be very useful.
uses custom hardware but might contain relevant bits. have not been able
to find a pdf though apart from through ie^3-xplore.
Radio localization in OFDM networks using the “round trip phase”
Ohlemueller, T. Winkler, F. Grass, E.
Humboldt Univ. Berlin, Berlin, Germany
2010 7th Workshop on Positioning Navigation and Communication (WPNC)
–
Sent from my fingertip …