Interference Detection and Localisation within GEMS II Ediz Cetin, Ryan J. R. Thompson and Andrew G. Dempster
GNSS Environmental Monitoring System (GEMS) ARC Linkage Project between: GEMS I : Comprehensively monitor GNSS system performance including operational failures and fault prediction, atmospheric influences, sudden GNSS satellite constellation changes and other system-critical information. GEMS II: Interference detection and localisation Consists of a number of spatially separated sensor nodes incorporating antenna arrays, connected to a master processing unit to quickly estimate the location of an interferer through hybrid Angle-of-Arrival (AOA) measurements at each node and Time- Difference-of-Arrival (TDOA) measurement between nodes Typical GEMS II Set-up 2
GEMS II Interference Detection Using typical GPS receiver observables to detect interference: Automatic Gain Control (AGC) and Carrier to Noise Ratio (C/No) AGC is an important element in the RF Front-end of a GNSS receiver used to minimize quantisation losses IF input VGA AGC gain voltage + σ σ IF σ IDEAL In the presence of RFI the AGC will reduce the gain from a nominal value in ADC +L -L +3 +1-1 -3 2-bit front-end ideal (L/σ=1).6dB loss order to main the desired saturation level, which can be used to detect its 4.2 presence. AGC control voltage (v) 3 4 3.8 3.6 RFI off 3.4 1 2 3 4 5 Time (1ms) RFI on threshold
GEMS II Interference Detection : AGC The response of the AGC varies with ambient temperature making detection of weaker interference more difficult. The gain/temperature coefficient of the receiver can be modelled over time, allowing for detection of weaker interference (~down to J/N of -8dB). AGC PulseWidth 46 455 45.5 1 1.5 2 2.5 3 Time (s) x 1 5 AGC PulseWidth 46 455 45 Temperature (C) 25 2 15 1 1 2 3 Time (s) x 1 5 AGC and temperature over couple of days 1 15 2 25 Temperature (C) Linear model of temperature versus AGC For test receiver: Δgt=-.776dB/deg. C 4
GEMS II Interference Detection: C/No C/No, which is available in the NMEA and RINEX standards, can be used for interference detection C/No can be a noisy measurement due to multipath, even for a stationary receiver in a calm environment (e.g. in IGS/CORS network) If the surrounding environment around the antenna does not change, the variations in C/No due to multipath repeat with satellites ground track. By tracking the difference of C/No measurements between ground tracks the effects of multipath can be reduced, especially at lower elevations C N eff = N C + N RFI C/No (db/hz) 45 44 43 42 41 4 39 38 C/No n (t) C/No n-1 (t) 1.7 1.8 1.9 2 2.1 UTC Time (s) x 1 4 C/No (db/hz) 45 44 43 42 41 4 39 38 C/No n (t+r) C/No n-1 (t) 1.7 1.8 1.9 2 2.1 UTC Time (s) x 1 4 C/No 3 2 1-1 -2-3 1.7 1.8 1.9 2 2.1 UTC Time(s) x 1 4 1. sync timestamps 2. difference 5
GEMS II TDOA Aspects TDOA based approach is one of the most commonly used techniques for locating emitters Received signals, and, at each receiver can be expressed as: where τ d is the time difference of arrival between the two received signals TDOA is estimated by maximizing the cross-correlation of signals received at two spatially separated receivers: 6
GEMS II TDOA Aspects TDOA based approach is one of the most commonly used techniques for locating emitters Received signals, and, at each receiver can be expressed as: where τ d is the time difference of arrival between the two received signals TDOA is estimated by maximizing the cross-correlation of signals received at two spatially separated receivers: 7
GEMS II TDOA Performance Analysis Simulation Set-up: - Sampling Frequency: 5.714 MHz - IF Frequency: 1.45 MHz - An area of 5x3km is sub-divided into cells of 1x1m resulting in 15 interference locations - Interfering Signal: White Gaussian noise with similar bandwidth to the GPS signal. - JNR level varying between -1 to 1 db -1 Monte Carlo simulations per interference location FD-ZP PB-CPSD PB-FD-CC CPSD-PH-WD CPSD-PH-WD-MG CPSD-PH-WD-3dB CPSD-PH-MG Histogram of Time-delay Estimate Error (ns) between R 1 and R 2, t 12, JNR= db 8
GEMS II TDOA Performance Analysis TDOA MSE (ns) for varying JNR and varying cross-correlation window length Mean Square Error (ns) 2 15 1 5-8 db -6 db -4 db db 4 db 6 db 8 db 1 db 1ms 2ms 3ms 4ms 5ms 6ms 1ms Cross-correlation Window Length (ms) TDOA MSE (ns) for varying interference bandwidth and JNR 9
GEMS II TDOA Performance Analysis Position Estimation Accuracy : Green <1m, Yellow 1 15m, Red >15m, JNR= db FD-ZP PB-CPSD PB-FD-CC CPSD-PH-WD CPSD-PH-WD-MG CPSD-PH-WD-3dB CPSD-PH-MG 1
GEMS II Dealing With Weak Interference: AOA Background noise subtraction allows very weak interferences to be detected: Background Noise changes slowly with time Detection threshold reduced from SNR of 7 db to SNR of -2 db 1 Signal + Noise Noise 5 After subtraction highest peak is at correct AOA. Beampowers (signal) Beampowers (Noise1) Beampowers (Noise2) Beampowers (Signal - Noise2) Power (db) -5-1 SNR = -18 db (estimated) Signal Direction -15-2 Signal Noise -25 5 1 15 2 25 3 35 4 Azimuth Angle (degree) 11
GEMS II Dealing With Weak Interference: TDOA Where is the interference? Cross-correlation peaks due to the GPS signals themselves limits the level at which actual interferers can be detected and localized since these unwanted peaks could be mistaken for weak interferers This problem is expected to be further exacerbated by the current and proposed deployment of a number of new Global Navigation Satellite Systems (GNSS) and regional augmentations. 12
GEMS II Dealing With Weak Interference: TDOA Sub-space projections to remove GPS signals to extend GEMS II sensitivity and coverage area. Cross-Correlation 3.5 3 2.5 2 1.5 1.5 4 x 14 Before Projections After 1 iteration After 3 iterations PRN8 PRN2 PRN1 PRN4 PRN28-4 -3-2 -1 1 2 3 4 Delay (Samples) PRN7 PRN25 PRN3 Interferer PRN6 PRN13 PRN23 PRN19 Max. cc-peak due to GPS signals before projections Interference (detected after 1 iteration) Max. cc-peak due to GPS signals after 1 iteration Max. cc-peak due to GPS signals after 5 iterations 13
GEMS II Dealing With Weak Interference: TDOA Average Mitigation and Detection for JNR values of -18, -2 and -22 db (Averaged across 1 Spirent based experiments) 8 8 Average Mitigation (%) 75 7 65 6-18JNR -2JNR -22JNR Average Mitigation (%) 75 7 65 6-18JNR -2JNR -22JNR 55 1 2 3 4 5 Iterations 1 55 1 2 3 4 5 Iterations 1 Average Detection (%) 8 6 4 2-18JNR -2JNR -22JNR Average Detection (%) 8 6 4 2-18JNR -2JNR -22JNR Initial 1 2 3 4 5 Iterations Baseline: 1m Initial 1 2 3 4 5 Iterations Baseline: 5m 14
Concluding Remarks Vulnerability of GNSS is an increasingly important problem as critical infrastructure is becoming ever more reliant on it to function. Detection, localisation and elimination of interference to the GPS has become of paramount importance. Briefly detailed GEMS II system developed for detecting and localising interference. Detailed various interference detection and localisation approaches and evaluated their performance. 15