Integrated Techniques for Interference Source Localisation in the GNSS band Joon Wayn Cheong Ediz Cetin Andrew Dempster
Introduction • GNSS signals are • A network of sensors inherently weak tuned to the GNSS band can be used to • Spurious detect the angle of transmissions and arrival (AOA) and time intentional jammers in difference of arrival the GNSS band (TDOA) of the jammer. threatens safety critical applications that depends on GNSS | 2 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Introduction • AOA: Uses phased antenna arrays DSP • TDOA: Uses cross- correlation method DSP • Geo-localisation of jammer – AOA: Intersection of lines – TDOA: Intersection of hyperbolas – Can we combine AOA and TDOA for Geo-localisation? | 3 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Jammer Characteristics • Narrowband – Strong jammer signal strength will affect receiver performance – Can be detected using AOA • Wideband – Weak jammer signal strength is sufficient to affect receiver performance – Can be detected using TDOA and AOA | 4 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Cramer Rao Bound: AOA AOA Σ � ∈ � ��� Measurement Covariance � � � � � � � � � � � � � � � � � ← ⋮ ⋮ Jacobian � � � � � � � � � � � � � � � �� � Σ � �� � � � � CRB • Most of the errors are within 10-40m • Errors behave smoothly outside the convex area | 5 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Cramer Rao Bound: TDOA TDOA Measurement Covariance: Σ � ∈ � ������� � � � � � � � � � � � � � � � � � � � � � � � � � � � � ← ⋮ ⋮ � � � � � � � � � � � � � � � � � � � � � � � � � � � Σ � �� � � �� � � CRB: • Most of the errors are within 5-40m • Errors behave erratically due to rank deficiency beyond the convex area bounded by the 3 nodes | 6 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
CRB for AOA + TDOA Integration • Most of the errors are within 2-30m • Rank deficient regions significantly improved • Lowest CRLB achieved at all points | 7 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Fair comparison between independent localisation and integrated localisation 1500 1 1000 500 0 2 3 -500 -1000 -1500 In Convex of 2 SN -2000 In Coverage of all 3 SN In Convex -2500 -2500 -2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 AOA STD: 0.5 degrees TDOA STD: 11m | 8 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
1500 Improvement over AOA-only 1 1000 500 0 2 3 -500 -1000 Improvement • -1500 In Convex of 2 SN -2000 In Coverage of all 3 SN measured in In Convex -2500 -2500 -2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 percentage (%) | 9 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
1500 Improvement over TDOA-only 1 1000 500 0 2 3 -500 -1000 Improvement • -1500 In Convex of 2 SN -2000 In Coverage of all 3 SN measured in In Convex -2500 -2500 -2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 percentage (%) | 10 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
AOA + TDOA Fusion Architectures Loose Integration Tight Integration | 11 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Loose Integration Algorithm Compute Position Error Covariance Matrix for � � , � � Input : AOA measurements � � , ∀� ∈ 1, … , � � � � �� � � � �� � TDOA measurements �̂ �� , ∀� ∈ 2, … , � � � � � � � Sensor Node Positions � � , � � , ∀� ∈ 1, … , � � � ← ⋮ ⋮ � � � �� � � � �� � Source Guesstimate Position � � , � � � � � � � � AOA Noise Covariance Matrix Σ � ∈ � ��� �� �� � � � Σ � Σ � ← � � TDOA Noise Covariance Matrix Σ � ∈ � ������� Perform Loose Integration Σ � 0 ��� Output : � �, � � Estimated Emitter Position Σ � 0 ��� Σ � Initialise � �, � � ← � � , � � � � ← � ��� Compute TDOA-only solution with arguments: �̂ �� , � � , � � , Σ � � ��� � � Output stored as � � , � � � � � � Compute AOA-only solution with arguments: � � , � � , � � , Σ � � Σ �� � � �� � � � Σ �� � ← � � � � � Output stored as � � , � � � � Compute Position Error Covariance Matrix for � � , � � � � � � � � � � � ← � � � � �� � � � � �� � � � �� � � � � �� � � � � � � � � � � � ← ⋮ ⋮ � � �� � � � � �� � � � �� � � � � �� � � � � � � � � � � Σ � �� � � �� Σ � ← � � | 12 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Loose Integration Algorithm Compute Position Error Covariance Matrix for � � , � � Input : AOA measurements � � , ∀� ∈ 1, … , � � � � �� � � � �� � TDOA measurements �̂ �� , ∀� ∈ 2, … , � � � � � � � Sensor Node Positions � � , � � , ∀� ∈ 1, … , � � � ← ⋮ ⋮ � � � �� � � � �� � Source Guesstimate Position � � , � � � � � � � � AOA Noise Covariance Matrix Σ � ∈ � ��� �� �� � � � Σ � Σ � ← � � TDOA Noise Covariance Matrix Σ � ∈ � ������� Perform Loose Integration Σ � 0 ��� Output : � �, � � Estimated Emitter Position Σ � 0 ��� Σ � Initialise � �, � � ← � � , � � � � ← � ��� Compute TDOA-only solution with arguments: �̂ �� , � � , � � , Σ � � ��� � � Output stored as � � , � � � � � � Compute AOA-only solution with arguments: � � , � � , � � , Σ � � Σ �� � � �� � � � Σ �� � ← � � � � � Output stored as � � , � � � � Compute Position Error Covariance Matrix for � � , � � � � � � � � � � � ← • The key to loose integration is the � � computation of an accurate Position Error � � �� � � � � �� � � � �� � � � � �� � � � � � � � � � Covariance Matrix for AOA and TDOA � � ← ⋮ ⋮ � � �� � � � � �� � � � �� � � � � �� � systems. � � � � � � � � • requires an approximate position to be � Σ � �� � � �� Σ � ← � � provided • provides a “weighing” mechanism | 13 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Effect of incorrect weighing matrix Correct Weighing Incorrect Weighing | 14 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Tight Integration Algorithm � � � � � � � � � Input : AOA measurements � � , ∀� ∈ 1, … , � � � � � � � ⋮ ⋮ TDOA measurements �̂ �� , ∀� ∈ 2, … , � � � � � � � � � � � � � � Sensor Node Positions � � , � � , ∀� ∈ 1, … , � � � � � ← � � � � � � � � � � � � � � � � � � AOA Noise Covariance Matrix Σ � ∈ � ��� � � � � � � � � ⋮ ⋮ TDOA Noise Covariance Matrix Σ � ∈ � ������� � � � � � � � � � � � � � � � � � � � � � � Output : � �, � � Estimated Emitter Position � � � � Iterate: Δ� � �, � ← � �, � � ⋮ � � � � � Δ� Δ� � � � ← � Σ �� � � �� � � � Σ �� Δ� ← � � � � � �� ⋮ � �� Δ� � � � � � ← Δ� � � End | 15 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Conclusion • AOA and TDOA Integration provides superior performance under all circumstances • Existing attempts to combine AOA and TDOA has been suboptimal due to incorrect “weighing” and/or use of a Loose Integration Architecture • 2 architectures has been proposed that can be adapted to various existing platforms • Proposed algorithms of both architectures approaches the Cramer Rao Lower Bound | 16 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Acknowledgement • GPSat Systems Australia • Ryan Thompson | 17 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Questions? Email: cjwayn@unsw.edu.au | 18 IGNSS 2016 - UNSW Sydney Australia – 6-8 December 2016
Recommend
More recommend