Sensor Selection For Mitigation of RSSBased Attacks in Wireless Local Area Network Positioning

1
Sensor Selection For Mitigation of RSS-Based
Attacks in Wireless Local Area Network Positioning

• Azadeh Kushki, Konstantinos N. Plataniotis,
  Anastasios N. Venetsanopoulos
• Speaker Hau-Ru Huang

2
System Model
3
System overview.
4
Attack Scenarios

• Fingerprinting-based positioning systems are
  vulnerable to malicious attacks (impersonation,
  jamming, and modifying signal strength).
• the proposed method selects a set of reliable APs
  from the set of available APs to perform
  positioning.
• The attack model
• We perturb the received power by adding Gaussian
  noise with variance

5
Spatial Filtering

• Coverage vector
• the Hamming distance
  between each anchor points coverage vector and
  that of the observation.

6
Position Estimation

• The fingerprinting set after spatial filtering
• The mean RSS form the a-th AP at Pi
• Given an observation from AP a, the MMSE
  estimate of position P is the posterior mean
  .

7
Position Estimation

• where KH() is the d-dimensional Gaussian kernel
  defined as

8
Position Estimation

• The resulting density can be interpreted as a
  Gaussian mixture

9
Position Estimation
10
Sensor Selection

• is the estimation erroe associated
  with the a-th AP.
• The small the covariance, the higher the
  probability that the estimator is close to the
  estimation mean
• Selecting a set of APs with small estimation.
  

11
Experimental setup
12
Result (1)

• Nlt0.2N all survey points are excluded and no
  estimate is available.

13
Result (2)

• N0.25N

14
Conclusion

• Experimental result indicate graceful performance
  degradation even when RSS measurements from all
  APs are corrupted.
• Resilience to RSS-based attacks was achieved
  through the selection of APs used in positioning
  based on a nonparametric estimate.