Time-domain beam signals for adaptive beamforming UDT 2019, Stockholm - 13th May … a sound decision … a sound decsion The ATLAS ELEKTRONIK Group/ 1
Table of contents • Motivation • For adaptive beamforming • ATLAS Approach • Robust MPDR (Minimum Power Distortionless Response) • Robust MPDR with time-domain signals • Simulated data • Sea trial data • Summary The ATLAS ELEKTRONIK Group/ 2
Table of contents • Motivation • For adaptive beamforming • ATLAS Approach • Robust MPDR (Minimum Power Distortionless Response) • Robust MPDR with time-domain signals • Simulated data • Sea trial data • Summary The ATLAS ELEKTRONIK Group/ 3
Motivation Basics of Beamforming • Joint processing of array outputs: • Enhanced detection of weak signals • Spatial discrimination of wave fronts • Spatial sensitivity depends on: • Array geometry • Frequency • Desired look-direction • Beamformer (BF) beamformer time delay … Δ𝜐 𝑂 Δ𝜐 1 Δ𝜐 2 • Example: „Delay & Sum“ Beamformer • BF coefficients = time delays Δ𝜐 𝑜 ∑ • Choice of Δ𝜐 𝑜 according to desired look-direction beam time series The ATLAS ELEKTRONIK Group/ 4
Motivation Beamforming Characteristics Beampattern Incoming wave front • Bearing = 90 ° • modulated by signal 𝑡(𝑢) Desired look-direction = 90 ° 𝑡(𝑢) beam time series Ƹ The ATLAS ELEKTRONIK Group/ 5
Motivation Beampattern – Mainlobe Mainlobe gain target detection Mainlobe width target separation beam time series Ƹ 𝑡(𝑢) The ATLAS ELEKTRONIK Group/ 6
Motivation Beampattern – Sidelobes Contains all acoustic signatures „collected“ with the beampattern! beam time series Ƹ 𝑡(𝑢) The ATLAS ELEKTRONIK Group/ 7
Motivation Disadvantage of non -adaptive Beamformers Contributions of loud target received via sidelobe @ 60 ° … 𝑡(𝑢) …mask contributions of silent target in Ƹ silent target might not be detected! …are erroneously assigned to bearing of 45 ° Loud target incorrect bearing for loud target! Silent target silent target + loud target + noise 𝑡(𝑢) beam time series Ƹ The ATLAS ELEKTRONIK Group/ 8
Motivation Idea of adaptive Beamforming (ABF) • Adapt beamformer coefficients such that directional zeros are formed in directions of interferers. • Derive information about interferers from data itself. Loud target • Perform adaptation of BF coefficients for each desired look-direction. Silent target silent target + noise 𝑡(𝑢) beam time series Ƹ The ATLAS ELEKTRONIK Group/ 9
Table of contents • Motivation • For adaptive beamforming • ATLAS Approach • Robust MPDR • Robust MPDR with time-domain signals • Simulated data • Sea trial data • Summary The ATLAS ELEKTRONIK Group/ 10
ATLAS Approach Robust Minimum Power Distortionless Response (MPDR) Basic idea of MPDR • Select BF coefficients such that: • Signal from look-direction remains undistorted • Total power of beam time series is minimized • Required information: • Covariance matrix of stave data (correlation of stave outputs) • Robust design of processing: • Introduce tolerance regions such that signals from a sector around look- direction remain undistorted prohibits suppression of targets between two look-directions • Calculate output power directly from steering vectors: • Matrix with dimension #beams x #frequency bands • Low time resolution (~ 1 Hz) The ATLAS ELEKTRONIK Group/ 11
ATLAS Approach Robust Minimum Power Distortionless Response (MPDR) Current Version • ABF for BDT processing • Conventional beamforming for BDT processing and others ABF Antenna-Lan Adaptive Covariance steering BDT Power matrix vectors ABF Conventional BDT beamforming e.g. LOFAR The ATLAS ELEKTRONIK Group/ 12
ATLAS Approach Robust Minimum Power Distortionless Response (MPDR) New design • ABF for BDT processing • Adaptive time-domain beamforming for BDT processing and others ABF Antenna-Lan Adaptive Covariance steering BDT Power matrix vectors ABF Adaptive time-domain BDT beamforming e.g. LOFAR The ATLAS ELEKTRONIK Group/ 13
Table of contents • Motivation • For adaptive beamforming • ATLAS Approach • Robust MPDR • Robust MPDR with time-domain signals • Simulated data • Simple scenario • LOFAR / DEMON results • Sea trial data • Summary The ATLAS ELEKTRONIK Group/ 14
Simulated data Simple scenario for a flank array sonar Scenario: • 4 broadband targets • Each has a strong frequency line Conventional Beamforming: • Target width depends on frequency • Sidelobes due to broadband signature • Strong sidelobe structure due to frequency lines The ATLAS ELEKTRONIK Group/ 15
Simulated data Simple scenario for a flank array sonar Scenario: • 4 broadband targets • Each has a strong frequency line ABF Beamforming: • Constant power width for broad frequency range • No sidelobes for broadband structure • No sidelobes for frequency lines • Improved performance • No time signals The ATLAS ELEKTRONIK Group/ 16
Simulated data Simple scenario for a flank array sonar Scenario: • 4 broadband targets • Each has a strong frequency line ABF Beamforming with time signals: • Constant power width for broad frequency range • No sidelobes for broadband structure • No sidelobes for frequency lines • Nearly same performance as before • Time Signals are available The ATLAS ELEKTRONIK Group/ 17
Simulated data Low Frequency Analysis and Recording (LOFAR) Intention : Analysis of frequency lines Engines Generators 𝑔 CFR : Cylinder firing rate PD 𝑔 EFR : Engine firing rate Pumps 𝑂 C : Number of cylinders 𝑔 𝑔 𝑔 EFR = 𝑔 CFR ∙ 𝑂 C Signal Processing : CFR LOFAR Frequency Decimation Normalization analysis The ATLAS ELEKTRONIK Group/ 18
Simulated data LOFAR (bearing information) Delay-and-Sum 5 simulated targets • Flank Array Sonar • Multiple target crossings The ATLAS ELEKTRONIK Group/ 19
Simulated data LOFAR (bearing information) Delay-and-Sum ABF Improved detection performance: 5 simulated targets • • Flank Array Sonar Improved target separation • Multiple target crossings The ATLAS ELEKTRONIK Group/ 20
Simulated data LOFAR (frequency information) Delay-and-Sum Simulation of frequency lines • Different SNR • Stable / unstable lines The ATLAS ELEKTRONIK Group/ 21
Simulated data LOFAR (frequency information) Delay-and-Sum ABF Improved detection performance: Simulation of frequency lines • • Different SNR Higher signal-to-noise-plus-interference ratio • • Stable / unstable lines Detection of more frequency lines possible The ATLAS ELEKTRONIK Group/ 22
Simulated data Detection of Envelope Modulation on Noise (DEMON) Intention : Analysis of frequency lines from modulation Cavitation 𝑔 PSR : Propeller shaft rate PD Bubbles generated by 𝑔 BR : Blade rate propeller 𝑂 B : Number of blades Signal processing : 𝑔 𝑔 𝑔 BR = 𝑔 PSR ∙ 𝑂 B PSR DEMON Absolute value LOFAR The ATLAS ELEKTRONIK Group/ 23
Simulated data DEMON Delay-and-Sum 5 simulated targets • Flank Array Sonar • Multiple target crossings The ATLAS ELEKTRONIK Group/ 24
Simulated data DEMON Delay-and-Sum ABF Improved detection performance: 5 simulated targets • • Flank Array Sonar Superior target separation • Multiple target crossings The ATLAS ELEKTRONIK Group/ 25
Table of contents • Motivation • For adaptive beamforming • ATLAS Approach • Robust MPDR • Robust MPDR with time-domain signals • Simulated data • Sea trial data • Summary The ATLAS ELEKTRONIK Group/ 26
Sea trial data Broadband Detection (BDT) Delay-and-Sum ≥ 14 target traces • Flank Array Sonar • 360 ° turn of the submarine • Reduced performance in endfire The ATLAS ELEKTRONIK Group/ 27
Sea trial data Broadband Detection (BDT) Delay-and-Sum Endfire ≥ 14 target traces • Flank Array Sonar • 360 ° turn of the submarine • Reduced performance in endfire The ATLAS ELEKTRONIK Group/ 28
Sea trial data Broadband Detection (BDT) Delay-and-Sum ABF Endfire ≥ 14 target traces Improved detection performance: • • Flank Array Sonar Higher signal-to-noise-plus-interference ratio • • 360 ° turn of the submarine Improved target separation • Reduced performance in endfire The ATLAS ELEKTRONIK Group/ 29
Sea trial data Broadband Detection (BDT) Delay-and-Sum ABF Endfire ≥ 14 target traces Improved detection performance: • • Flank Array Sonar Higher signal-to-noise-plus-interference ratio • • 360 ° turn of the submarine Improved target separation • Reduced performance in endfire The ATLAS ELEKTRONIK Group/ 30
Sea trial data Low Frequency Analysis and Recording (LOFAR) (Maximum from frequeny domain) Delay-and-Sum ≥ 14 target traces • Flank Array Sonar • 360 ° turn of the submarine • Reduced performance in endfire The ATLAS ELEKTRONIK Group/ 31
Recommend
More recommend
