A Hybrid TOARSS Based Location Estimation

1
A Hybrid TOA/RSS Based Location Estimation

• Zafer Sahinoglu, zafer_at_merl.com
• Digital Communications and Networking Group, MTL
• September 11th, 2004

2
Outline

• Hybrid ranging observation scenarios
• Modeling of RSS and TOA observations
• Problem formulation and Derivation of CRBs
• Results
• Summary and Conclusions

For More Details

• Proc. IEEE ICC 2004, June 2004, Paris
• IEEE Communications Letters, to appear in October
  2004

3
Hybrid Ranging Observation Scenarios
4
Wideband Channel Measurement Experiment PATWARI

• Office area partitioned by 1.8m high cubicle
  walls
• DSSS Tx and Rx (Sigtek model ST-515)
• 40MHz chip rate, fc 2.443GHz
• Omni-directional antennas 1m above the floor
• The Rx
• Down converts and correlates I and Q samples with
  the known PN signal and outputs a power-delay
  profile (PDP)
• Samples of the leading edge of the PDP is
  compared to an over-sampled (120MHz) template of
  the auto-correlation of the PN
• SNRgt25dB

5
Modeling of TOA and RSS Observations

• RSS obscured by log-normal shadowing
• Frequency-selective fading reduced by wideband
  average
• Time-averaging reduce fading due to motion of
  objects in channel, reciprocal channel averaging
  helps to reduce device calibration errors
• Log-normal shadowing remains
• TOA is affected predominantly by multipath
• Positive bias due to multipath assumed known and
  subtracted
• Resulting statistic

6
Relative Location Estimation Problem

• 1 sensor device (SN)
• m TOA devices with indexes 1,,m
• n RSS devices with indexes m1,mn
• Estimate the actual coordinate
• TOA observation Ti,j , time delay between
  devices i and j
• RSS observation Pi,j , received power at
  device j from i
• The estimation is based on (m-1) TOA and n RSS
  observations in the TDOA/RSS case
• Observation vector
• X XT XR T1,2, T2,3,Tm-1,m P0,m1,,
  P0,mn , TDOA/RSS hybrid scheme
• X XT XR T1, T2,Tm P0,m1,, P0,mn ,
  TOA/RSS hybrid scheme

7
Motivation for the Cramer-Rao Study

• The CRB provides the lower bound on the
  covariance matrix of any unbiased estimator
• Theoretical confirmation of whether a given
  scheme can satisfy applications precision ranging
  requirements
• Quantification of how random system/environment
  variables affect the precision ranging
• Useful for selection and optimization of design
  parameters
• The CRB of any unbiased estimator is
• is the Fisher Information Matrix
• The log-likelihood function is

8
Definition Geometric Conditioning

• Illustration of the geometric conditioning (A1,2)
  of devices 1 and 2 with respect to device
  0.

2
A
D
d0x1,2
1
0
C
B
9
Derivation of the CRB in TOA/RSS

• The CRB

TOA/RSS contribution
RSS contribution
TOA contribution
where and
10
The CRB for TOA vs TOA/RSS

• Four reference devices at four corners,
  separation 18m
• RSS suppresses singularities of TOA at corners
• Figures below gives in meters

11
Derivation of the CRB in TDOA/RSS

• The variance of the TDOA observations are twice
  higher than the TOA
• 1 TOA measurement is sacrificed for offset
  removal
• The CRB must therefore be higher than the TOA/RSS
• Geometric conditioning of APRs with respect to
  RNs directly affect the bound

1 index of the reference APR
TDOA contribution
TDOA/RSS contribution
RSS contribution
12
The CRB for TOA/RSS vs TDOA/RSS

• In TDOA/RSS, one reference device placed in the
  center, the other three around a circle to
  maintain a symmetric plot
• The radius of the circle is selected such that
  the area would be equal to 18x18m square
• TDOA/RSS inferior due to sacrificing 1
  independent TOA measurement and increased
  standard deviation
• The plots show the bounds in meters (np 2.3)

13
The Spatial Average of the CRBs
Spatial mean of the lower bound in meters
APR separation around the square in meters
14
Summary and Discussion

• The RSS measurements can be used to refine
  wideband TOA based estimations in short ranges
• The hybrid schemes TDOA/RSS and TOA/RSS
  outperform TOA or RSS based schemes alone

15
References

• PATWARI N. Patwari, A. O. Hero, M. Perkins, N.
  S. Correal, R. J. ODea Relative Location
  Estimation in Wireless Sensor Networks, IEEE
  Trans. Signal Processing, vol. 51, pp. 2137-2148,
  August 2003
• CAPKUN S. Capkun, M. Hamdi, and J.-P. Hubaux.
  GPS-free positioning in mobile ad-hoc network. In
  34th IEEE Hawaii International Conference on
  System Sciences (HICSS-34), Jan. 2001
• DOHERTY L. Doherty, K. S. pister, and L. E.
  Ghaoui. Convex position estimation in wireless
  sensor networks. In IEEE INFOCOM, vol. 3, pages
  1655 1663, 2001.
• ALBOVITCZ J. Albowicz, A. Chen, and L. Zhang.
  Recursive position estimation in sensor networks.
  In IEEE International Conference on Network
  Protocols, pp. 3541, Nov 2001.