Introduction - Motivation - Fault Model - Measurement Setup Fault Detection and Mitigation in WLAN RSS Nearest Neighbor Fingerprint-based Positioning Algorithm - Fault Detection - Fault Tolerance Hybrid Positioning Algorithm Christos Laoudias , Michalis Michaelides and Christos Panayiotou Experimental Evaluation KIOS Research Center for Intelligent Systems and Networks - Results Department of Electrical and Computer Engineering Conclusions University of Cyprus, Nicosia, Cyprus - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Outline Introduction - Motivation Introduction - Fault Model - Measurement Setup Nearest Neighbor Nearest Neighbor Algorithm Algorithm - Fault Detection - Fault Tolerance Hybrid Positioning Algorithm Hybrid Positioning Algorithm Experimental Experimental Evaluation Evaluation - Results Conclusions Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Motivation of our work Main focus of fingerprint positioning algorithms has been on Introduction reducing the positioning error which ranges between 2-10m - Motivation depending on the - Fault Model - Measurement Setup Nearest Neighbor ◮ underlying method (deterministic, probabilistic, etc) Algorithm - Fault Detection ◮ experimentation parameters (number of fingerprints collected, - Fault Tolerance resolution of the reference locations, density of the APs) Hybrid Positioning Algorithm Experimental Evaluation - Results Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Motivation of our work Main focus of fingerprint positioning algorithms has been on Introduction reducing the positioning error which ranges between 2-10m - Motivation depending on the - Fault Model - Measurement Setup Nearest Neighbor ◮ underlying method (deterministic, probabilistic, etc) Algorithm - Fault Detection ◮ experimentation parameters (number of fingerprints collected, - Fault Tolerance resolution of the reference locations, density of the APs) Hybrid Positioning Algorithm Experimental Evaluation Fault Tolerance - Results It is desirable to provide smooth performance degradation in the Conclusions - Concluding Remarks presence of faults, due to unpredicted failures or malicious attacks. International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Motivation of our work Main focus of fingerprint positioning algorithms has been on Introduction reducing the positioning error which ranges between 2-10m - Motivation depending on the - Fault Model - Measurement Setup Nearest Neighbor ◮ underlying method (deterministic, probabilistic, etc) Algorithm - Fault Detection ◮ experimentation parameters (number of fingerprints collected, - Fault Tolerance resolution of the reference locations, density of the APs) Hybrid Positioning Algorithm Experimental Evaluation Fault Tolerance - Results It is desirable to provide smooth performance degradation in the Conclusions - Concluding Remarks presence of faults, due to unpredicted failures or malicious attacks. Assumption The RSS data collected in the offline phase is not corrupted and we focus on AP failures and non-cryptographic RSS attacks that may occur during positioning. International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
AP Failure model Effect Introduction ◮ APs detected in the offline phase are not available during - Motivation - Fault Model positioning - Measurement Setup Nearest Neighbor Feasibility Algorithm - Fault Detection ◮ Unpredicted AP failures, e.g. power outage, WLAN system - Fault Tolerance Hybrid Positioning maintenance, AP firmware upgrade etc Algorithm ◮ AP shut down temporarily or removed permanently (public Experimental Evaluation WLAN systems) - Results Conclusions ◮ Adversary cuts off the power supply or severely jams the - Concluding Remarks communication channel Simulation ◮ Remove the RSS values of the faulty AP in the original test fingerprints International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Measurement Setup ◮ Area 560m 2 at KIOS Introduction Research Center, Cyprus - Motivation - Fault Model ◮ 73 WLAN APs (9 local, 64 - Measurement Setup neighboring) Nearest Neighbor Algorithm ◮ HP iPAQ hw6915 PDA - Fault Detection - Fault Tolerance Hybrid Positioning Training data Algorithm ◮ 105 reference locations, 40 Experimental Evaluation fingerprints per location - Results (4200 in total) Conclusions - Concluding Remarks Testing data ◮ 96 test locations, 20 fingerprints per location (1920 in total) International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Nearest Neighbor Algorithm Introduction - Motivation - Fault Model - Measurement Setup Nearest Neighbor Location Estimation Algorithm - Fault Detection � - Fault Tolerance � n � � � � 2 Hybrid Positioning � � ℓ ( s ) = arg min D i = r ij − s j ℓ i ∈ L D i , Algorithm Experimental j =1 Evaluation - Results Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Fault Detection Main Idea ◮ Exploit the distances D i that are already computed to decide Introduction whether fingerprint s is corrupt or not - Motivation - Fault Model - Measurement Setup ◮ The value of a distance-based fault indicator will violate a Nearest Neighbor certain ’fault-free’ threshold Algorithm - Fault Detection - Fault Tolerance Hybrid Positioning Algorithm Experimental Evaluation - Results Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Fault Detection Main Idea ◮ Exploit the distances D i that are already computed to decide Introduction whether fingerprint s is corrupt or not - Motivation - Fault Model - Measurement Setup ◮ The value of a distance-based fault indicator will violate a Nearest Neighbor certain ’fault-free’ threshold Algorithm - Fault Detection - Fault Tolerance Proposed Fault Indicator Hybrid Positioning Algorithm ◮ Sum of distances to the K nearest neighbors D ( K ) Experimental sum Evaluation - Results Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Fault Detection Main Idea ◮ Exploit the distances D i that are already computed to decide Introduction whether fingerprint s is corrupt or not - Motivation - Fault Model - Measurement Setup ◮ The value of a distance-based fault indicator will violate a Nearest Neighbor certain ’fault-free’ threshold Algorithm - Fault Detection - Fault Tolerance Proposed Fault Indicator Hybrid Positioning Algorithm ◮ Sum of distances to the K nearest neighbors D ( K ) Experimental sum Evaluation - Results Conclusions Fault Detection Steps - Concluding Remarks ◮ Select an appropriate threshold γ based on the distribution of the fault indicator D ( K ) sum in the fault-free case ◮ Fault is detected during positioning if D ( K ) sum > γ for the currently observed fingerprint International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Fault Detection in practice ◮ As the number of faulty APs is increased the CDF curve of D (2) sum is shifted to the right Introduction - Motivation - Fault Model - Measurement Setup Nearest Neighbor Algorithm - Fault Detection - Fault Tolerance Hybrid Positioning Algorithm Experimental Evaluation - Results Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Fault Detection in practice ◮ As the number of faulty APs is increased the CDF curve of D (2) sum is shifted to the right Introduction - Motivation - Fault Model - Measurement Setup Nearest Neighbor Algorithm - Fault Detection - Fault Tolerance Hybrid Positioning Algorithm Experimental Evaluation - Results Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Fault Detection in practice ◮ As the number of faulty APs is increased the CDF curve of D (2) sum is shifted to the right Introduction ◮ D (2) sum < 76 dBm for 95% of time, thus γ = 76 dBm (5% false - Motivation - Fault Model detections are acceptable) - Measurement Setup Nearest Neighbor Algorithm - Fault Detection - Fault Tolerance Hybrid Positioning Algorithm Experimental Evaluation - Results Conclusions - Concluding Remarks International Conference on Indoor Positioning and Indoor Navigation, Guimar˜ aes, Portugal 21 September 2011
Recommend
More recommend
