[Dataset] vanderbilt/interferometric (v. 2007-06-06)

top

the initial version

@MISC{vanderbilt-interferometric-2007-06-06,
  author = {Branislav Kusy and Janos Sallai},
  title = {{CRAWDAD} data set vanderbilt/interferometric (v. 2007-06-06)}, 
  howpublished = {Downloaded from http://crawdad.cs.dartmouth.edu/vanderbilt/interferometric},
  month = jun,  
  year = 2007
}
					

We collected localization traces from a radio interferometric tracking system, which is implemented on mote-class wireless sensor nodes.

Location-awareness is an important requirement for many
mobile wireless applications today. When GPS is not applicable
because of the required precision and/or the resource
constraints on the hardware platform, radio interferometric
ranging may offer an alternative. 

In [kusy-intereferometric], we present a technique that enables 
the precise tracking of multiple wireless nodes simultaneously. 
To evaluate the performance of the technique, we use this dataset
which was collected from a prototype implementation on mote-class 
wireless sensor nodes.

Our tracking system (called 'mTrack') is implemented using XSM motes
from Crossbow, Inc. (a variant of the Berkeley Mica2 mote) as target 
and infrastructure nodes, as well as a PC laptop running the location 
computation and the tracking GUI. The motes are running TinyOS version 
1.1.14 as the operating system. Our test environment is the empty 
Vanderbilt Football Stadium.

Sensors record the phase and frequency of the beat signal, 
while the transmitter pair iterates through a series of
transmit frequencies. 

Q-ranges are calculated for every receiver pair and then 
converted to t-ranges, which are input to the analytical 
location solver. Once positions are known, q-speeds are 
converted to velocity vectors. Positions and velocities 
are displayed on a map and optionally used to control 
camera pitch/pan.
(Please see [kusy-interferometric] for details about 
q-ranges, t-ranges, and q-speed.)

[Traceset] vanderbilt/interferometric/mtrack (v. 2007-06-06)

top

the initial version

@MISC{vanderbilt-interferometric-mtrack-2007-06-06,
  author = {Branislav Kusy and Janos Sallai},
  title = {{CRAWDAD} trace set vanderbilt/interferometric/mtrack (v. 2007-06-06)}, 
  howpublished = {Downloaded from http://crawdad.cs.dartmouth.edu/vanderbilt/interferometric/mtrack},
  month = jun,  
  year = 2007
}
					

Localization traceset collected from a radio interferometric tracking system which is implemented on Mica2 compatible XSM motes.

- Implementation

The radio-interferometric ranging engine is implemented
as an alternate radio driver. Since both the TinyOS radio
stack and the ranging engine use the same radio hardware,
the former is disabled during the measurement rounds. After
a measurement round is completed, the results (phase and 
frequency measurements for each frequency channel) are routed 
to the PC using the Directed Flood Routing Framework (DFRF). 
DFRF is configured with the gradient convergecast policy 
to provide a fast and reliable data collection service 
in a multihop network.

The computation of the interferometric q-ranges, their
conversion to t-ranges, as well as the location computation
are implemented in Java and run on a PC. (Please see 
[kusy-interferometric] for details about q-ranges and t-ranges.)
The computation is implemented in a dataflow like manner, which would
make it possible to distribute computational blocks to different
computers, or to implement them in hardware.

- Time Synchronization

To assure that the phase measurements are carried out at the same time 
on all receivers, the receivers need to be time synchronized. Instead of 
choosing a synchronization service that maintains a global time in the 
whole network (such as FTSP), we opted for a multihop extension of 
the Estimated Time on Arrival (ETA) approach: 
the transmitter node at the common focus of the hyperbolae, called 
the master node, generates a beacon event tagged with the start time 
of the next measurement round by the node's local clock. ETA will 
propagate this beacon message to all nodes within a few radio hops, 
converting the timestamp to the local time of the recipients.

ETA is able to achieve much better utilization of system
resources than virtual global time services, because it does
not synchronize the clock skews of different nodes. However,
the beacon event and the start of the measurement round need 
to be close in time (300 ms), to achieve high synchronization
accuracy despite the drifts of the unsynchronized local clocks 
at different nodes. Notice, that it is imperative that multihop 
time synchronization be used, because the maximum interferometric 
range exceeds the communication range of the motes.

- Experimental Environment 

Our experimental hardware platform is a Mica2 compatible
XSM mote, programmed using the nesC programming language and 
the TinyOS operating system. Our test environment is the empty 
Vanderbilt Football Stadium.

- Measurement Details

The mTrack system was able to get a position fix for each of 
the targets in approximately 4 seconds which includes
0.5 second coordination time, 1 second ranging time, 2 seconds 
multihop routing time, and 0.5 second localization time. 

During the coordination phase, infrastructure nodes were assigned 
the roles of transmitters and receivers and were time synchronized 
with each other and with the target nodes. The nodes measured the phase 
and frequency of the beat signal during the ranging phase and routed 
the measured values to the base station during the routing phase.
Finally, a PC computer calculated the target locations and velocity 
vectors during the localization phase. 

We estimated the accuracy of mTrack's localization and
velocity vector estimation with respect to the ground truth.
Measuring the ground truth locations of multiple moving
targets accurately, both spatially and temporally, however,
was a difficult problem. We simplified this problem by predefining
the tracks that the targets followed during the actual
experiment. Each track consisted of a series of waypoints
and the targets moved between two consecutive waypoints
at an approximately constant speed on a straight line. 

During the actual experiment, we recorded the times
at which each of the targets passed each of its predefined
waypoints. This allowed us to compute the actual speed of
the target for each segment of its predefined track. Moreover,
since we also recorded the times when the ranging measurements 
were taken, we could determine the segment and interpolate 
the ground truth location of the target on that segment for 
any given ranging measurement. Therefore, for any given ranging 
measurement, we were able to reconstruct the ground truth location 
of the target as well as its velocity vector.

- Experiment Scenario 

We deployed five anchor nodes at known surveyed locations, covering 
an area of approximately 27.4 - 27.4 m. We placed four anchors 
in the corners of this square and the fifth anchor close to the center. 
The actual setup can be seen in 
[Figure: the experimental setup for multipleMobile2hands(a)] and 
[Figure: the experimental setup for multipleMobile3], which shows 
the anchor node locations (as black dots), the track that a mobile 
node follows, and the calculated locations and velocity vectors. 

This dataset contains the data from the following three experiments:

1. multipleMobile2hands
Simultaneous tracking of multiple nodes: a person holds two motes 
in two hands, approximately 1.5 m apart and walks on the rectangular
track, second person holds a single mote and walks on the triangular track.
The setup is shown in [Figure: the experimental setup for multipleMobile2hands]. 

2. multipleMobile2handsa
Same as multipleMobile2hands, but at a different time.
The setup is shown in [Figure: the experimental setup for multipleMobile2handsa]. 

3. multipleMobile3
Simultaneous tracking of multiple nodes: three different nodes are 
moving along 3 different tracks, starting at points A,B, and C.
The setup is shown in [Figure: the experimental setup for multipleMobile3].

[Trace] vanderbilt/interferometric/mtrack/multipleMobile2hands (v. 2007-06-06)

top

the initial version

@MISC{vanderbilt-interferometric-mtrack-multipleMobile2hands-2007-06-06,
  author = {Branislav Kusy and Janos Sallai},
  title = {{CRAWDAD} trace vanderbilt/interferometric/mtrack/multipleMobile2hands (v. 2007-06-06)}, 
  howpublished = {Downloaded from http://crawdad.cs.dartmouth.edu/vanderbilt/interferometric/mtrack/multipleMobile2hands},
  month = jun,  
  year = 2007
}
					

Localization trace collected from a radio interferometric tracking system for tracking three mobile nodes.

Simultaneous tracking of multiple nodes: a person holds two motes 
in two hands, approximately 1.5 m apart and walks on the rectangular
track, second person holds a single mote and walks on the triangular track.
The setup is shown in [Figure: the experimental setup for multipleMobile2hands]. 

Interferometric measurement involves 3 nodes: master, assistant, and receiver. 
Master and assistant transmit at the same time to create the interference 
measurement and the receiver measures frequency and phase of this interference signal. 
This is repeated at multiple channels. In the paper [kusy-interferometric], 
we used the following 22 channels:

    chan-ID         1   2   3   4   5   6   7   8   9   10  11  12  13  14  15  16  17  18  19  20  21  22
    chan-freq(Mhz)  401.1409755 421.1528585 441.69137   417.9930875 438.531599  458.543482  402.1942325 422.2061155 442.744627  416.9398305 437.478342  457.490225  403.2474895 423.2593725 443.797884  415.8865735 436.425085  456.436968  404.3007465 424.3126295 444.851141  414.8333165

To measure the interferometric range (q-range), we need to take calculate phase offsets of two
receivers at all 22 channels.

.freq file:
    each line of this file stores the frequencies measured in one interference measurement, at one receiver node. 
	In the experiments in the paper [kusy-interferometric], only first 22 channels were measured, the values of the other channels are marked as NaN.
    parsing of each line: measurement-time   seqNum masterID   assistantID    receiverID freq@CH1 freq@CH2 freq@CH3  ...

    measurement-time is in hh:mm:ss.sss
    freq@CHx means frequency in Hz measured at the radio channel with chan-ID 'x'


.phase file:
    each line of this file stores the phases measured in one interference measurement, at one receiver node. 
	In the experiments in [kusy-interferometric] experiments, only first 22 channels were measured, the values of the other channels are marked as NaN.
    parsing of each line: measurement-time   seqNum masterID   assistantID    receiverID phase@CH1 phase@CH2 phase@CH3  ...

    measurement-time is in hh:mm:ss.sss
    phase@CHx means phase in radian measured at the radio channel with chan-ID 'x'

.pos file:
    each line of this file stores information about a sensor node.
    parsing of each line: 'Sensor' moteID   x   y   z   is-anchor   dontcare   dontcare   dontcare

    moteID is ID of the mote
    x is x-coordinate of the mote in meters
    y is y-coordinate of the mote in meters
    z is z-coordinate of the mote in meters
    is-anchor is one if the mote was anchor, zero if it was a tracked node

.track file:
    this file stores the ground truth track for each of the tracked nodes. it consists of multiple sections, each of them
    starting with #nodeID and continuing with the track data points: measurement-time   x   y   z
    each line should be interpreted as: at the measurement-time nodeID was located at (x,y,z)

.xls and .png files:
    these files show the output of our tracking algorithm (which takes frequencies, phases, and locations of anchor nodes
    and calculates interferometric ranges (q-ranges), q-speeds, t-ranges, and finally locations and velocities of the
    tracked nodes). excel file contains the data, png file shows the resulting figure published in the paper [kusy-interferometric].

.set file:
    this file stores various settings, most notably IDs of the channels at which we generated the interference signal.
    channelID can be converted to MHz the following way: 430.105543 MHz + channelID*0.5266285 MHz
    e.g. channelID1=-55 corresponds to 401.14 MHz

[Trace] vanderbilt/interferometric/mtrack/multipleMobile2handsa (v. 2007-06-06)

top

the initial version

@MISC{vanderbilt-interferometric-mtrack-multipleMobile2handsa-2007-06-06,
  author = {Branislav Kusy and Janos Sallai},
  title = {{CRAWDAD} trace vanderbilt/interferometric/mtrack/multipleMobile2handsa (v. 2007-06-06)}, 
  howpublished = {Downloaded from http://crawdad.cs.dartmouth.edu/vanderbilt/interferometric/mtrack/multipleMobile2handsa},
  month = jun,  
  year = 2007
}
					

Localization trace collected from a radio interferometric tracking system for tracking three mobile nodes.

Simultaneous tracking of multiple nodes: a person holds two motes 
in two hands, approximately 1.5 m apart and walks on the rectangular
track, second person holds a single mote and walks on the triangular track.
The experimetal setup is the same as multipleMobile2hands, but was conducted 
at a different time. 
The setup is shown in [Figure: the experimental setup for multipleMobile2handsa]. 

Interferometric measurement involves 3 nodes: master, assistant, and receiver. 
Master and assistant transmit at the same time to create the interference 
measurement and the receiver measures frequency and phase of this interference signal. 
This is repeated at multiple channels. In the paper [kusy-interferometric], 
we used the following 22 channels:

    chan-ID         1   2   3   4   5   6   7   8   9   10  11  12  13  14  15  16  17  18  19  20  21  22
    chan-freq(Mhz)  401.1409755 421.1528585 441.69137   417.9930875 438.531599  458.543482  402.1942325 422.2061155 442.744627  416.9398305 437.478342  457.490225  403.2474895 423.2593725 443.797884  415.8865735 436.425085  456.436968  404.3007465 424.3126295 444.851141  414.8333165

To measure the interferometric range (q-range), we need to take calculate phase offsets of two
receivers at all 22 channels.

.freq file:
    each line of this file stores the frequencies measured in one interference measurement, at one receiver node. 
	In the experiments in the paper [kusy-interferometric], only first 22 channels were measured, the values of the other channels are marked as NaN.
    parsing of each line: measurement-time   seqNum masterID   assistantID    receiverID freq@CH1 freq@CH2 freq@CH3  ...

    measurement-time is in hh:mm:ss.sss
    freq@CHx means frequency in Hz measured at the radio channel with chan-ID 'x'


.phase file:
    each line of this file stores the phases measured in one interference measurement, at one receiver node. 
	In the experiments in the paper [kusy-interferometric], only first 22 channels were measured, the values of the other channels are marked as NaN.
    parsing of each line: measurement-time   seqNum masterID   assistantID    receiverID phase@CH1 phase@CH2 phase@CH3  ...

    measurement-time is in hh:mm:ss.sss
    phase@CHx means phase in radian measured at the radio channel with chan-ID 'x'

.pos file:
    each line of this file stores information about a sensor node.
    parsing of each line: 'Sensor' moteID   x   y   z   is-anchor   dontcare   dontcare   dontcare

    moteID is ID of the mote
    x is x-coordinate of the mote in meters
    y is y-coordinate of the mote in meters
    z is z-coordinate of the mote in meters
    is-anchor is one if the mote was anchor, zero if it was a tracked node

.track file:
    this file stores the ground truth track for each of the tracked nodes. it consists of multiple sections, each of them
    starting with #nodeID and continuing with the track data points: measurement-time   x   y   z
    each line should be interpreted as: at the measurement-time nodeID was located at (x,y,z)

.xls and .png files:
    these files show the output of our tracking algorithm (which takes frequencies, phases, and locations of anchor nodes
    and calculates interferometric ranges (q-ranges), q-speeds, t-ranges, and finally locations and velocities of the
    tracked nodes). excel file contains the data, png file shows the resulting figure published in the paper [kusy-interferometric].

.set file:
    this file stores various settings, most notably IDs of the channels at which we generated the interference signal.
    channelID can be converted to MHz the following way: 430.105543 MHz + channelID*0.5266285 MHz
    e.g. channelID1=-55 corresponds to 401.14 MHz

[Trace] vanderbilt/interferometric/mtrack/multipleMobile3 (v. 2007-06-06)

top

the initial version

@MISC{vanderbilt-interferometric-mtrack-multipleMobile3-2007-06-06,
  author = {Branislav Kusy and Janos Sallai},
  title = {{CRAWDAD} trace vanderbilt/interferometric/mtrack/multipleMobile3 (v. 2007-06-06)}, 
  howpublished = {Downloaded from http://crawdad.cs.dartmouth.edu/vanderbilt/interferometric/mtrack/multipleMobile3},
  month = jun,  
  year = 2007
}
					

Localization trace collected from a radio interferometric tracking system for tracking three mobile nodes.

Simultaneous tracking of multiple nodes: three different nodes are 
moving along 3 different tracks, starting at points A,B, and C.
The setup is shown in [Figure: the experimental setup for multipleMobile3].   

Interferometric measurement involves 3 nodes: master, assistant, and receiver. 
Master and assistant transmit at the same time to create the interference 
measurement and the receiver measures frequency and phase of this interference signal. 
This is repeated at multiple channels. In the the [kusy-interferometric], 
we used the following 22 channels:

    chan-ID         1   2   3   4   5   6   7   8   9   10  11  12  13  14  15  16  17  18  19  20  21  22
    chan-freq(Mhz)  401.1409755 421.1528585 441.69137   417.9930875 438.531599  458.543482  402.1942325 422.2061155 442.744627  416.9398305 437.478342  457.490225  403.2474895 423.2593725 443.797884  415.8865735 436.425085  456.436968  404.3007465 424.3126295 444.851141  414.8333165

To measure the interferometric range (q-range), we need to take calculate phase offsets of two
receivers at all 22 channels.

.freq file:
    each line of this file stores the frequencies measured in one interference measurement, at one receiver node. 
	In the experiments in the paper [kusy-interferometric], only first 22 channels were measured, the values of the other channels are marked as NaN.
    parsing of each line: measurement-time   seqNum masterID   assistantID    receiverID freq@CH1 freq@CH2 freq@CH3  ...

    measurement-time is in hh:mm:ss.sss
    freq@CHx means frequency in Hz measured at the radio channel with chan-ID 'x'


.phase file:
    each line of this file stores the phases measured in one interference measurement, at one receiver node. 
	In the experiments in the paper [kusy-interferometric], only first 22 channels were measured, the values of the other channels are marked as NaN.
    parsing of each line: measurement-time   seqNum masterID   assistantID    receiverID phase@CH1 phase@CH2 phase@CH3  ...

    measurement-time is in hh:mm:ss.sss
    phase@CHx means phase in radian measured at the radio channel with chan-ID 'x'

.pos file:
    each line of this file stores information about a sensor node.
    parsing of each line: 'Sensor' moteID   x   y   z   is-anchor   dontcare   dontcare   dontcare

    moteID is ID of the mote
    x is x-coordinate of the mote in meters
    y is y-coordinate of the mote in meters
    z is z-coordinate of the mote in meters
    is-anchor is one if the mote was anchor, zero if it was a tracked node

.track file:
    this file stores the ground truth track for each of the tracked nodes. it consists of multiple sections, each of them
    starting with #nodeID and continuing with the track data points: measurement-time   x   y   z
    each line should be interpreted as: at the measurement-time nodeID was located at (x,y,z)

.xls and .png files:
    these files show the output of our tracking algorithm (which takes frequencies, phases, and locations of anchor nodes
    and calculates interferometric ranges (q-ranges), q-speeds, t-ranges, and finally locations and velocities of the
    tracked nodes). excel file contains the data, png file shows the resulting figure published in the paper [kusy-interferometric].

.set file:
    this file stores various settings, most notably IDs of the channels at which we generated the interference signal.
    channelID can be converted to MHz the following way: 430.105543 MHz + channelID*0.5266285 MHz
    e.g. channelID1=-55 corresponds to 401.14 MHz

[Author] Branislav Kusy

top

[Author] Janos Sallai

top

[Paper] kusy-interferometric

top

Location-awareness is an important requirement for many mobile wireless 
applications today. When GPS is not applicable because of the required 
precision and/or the resource constraints on the hardware platform, radio 
interferometric ranging may offer an alternative. In this paper, we present a 
technique that enables the precise tracking of multiple wireless nodes 
simultaneously. It relies on multiple infrastructure nodes deployed at known 
locations measuring the position of tracked mobile nodes using radio 
interferometry. In addition to location information, the approach also provides 
node velocity estimates by measuring the Doppler shift of the interference 
signal. The performance of the technique is evaluated using a prototype 
implementation on mote-class wireless sensor nodes. Finally, a possible 
application scenario of dirty bomb detection in a football stadium is briefly 
described.