MINISTRY OF EDUCATION AND TRAINING MINISTRY OF NATIONAL DEFENCE ACADEMY OF MILITARY SCIENCE AND TECHNOLOGY TRAN CONG TRANG THE RESEARCH OF IMPROVING THE TARGET POSITIONING ACCURACY FOR
Trang 1MINISTRY OF EDUCATION AND TRAINING MINISTRY OF NATIONAL DEFENCE
ACADEMY OF MILITARY SCIENCE AND TECHNOLOGY
TRAN CONG TRANG
THE RESEARCH OF IMPROVING THE TARGET POSITIONING ACCURACY FOR SHIP’S PASSIVE SONAR SYSTEMS
Specialization: Radar Technology and Navigation
Code: 9 52 02 04
SUMMARY OF Ph.D THESIS IN ENGINEERING
Hanoi, 2019
Trang 2This thesis has been completed at:
ACADEMY OF MILITARY SCIENCE AND TECHNOLOGY
Scientific supervisors:
1 Dr Tran Van Hung
2 Dr Nguyen Thanh Hung
Reviewer 1: Assoc Prof, Dr Vu Van Yem
Hanoi University of Science and Technology
Reviewer 2: Assoc Prof, Dr Le Vinh Ha
Academy of Military Science and Technology
Reviewer 3: Dr Nguyen Van So
189 Brigade, Vietnam People’s Navy
The thesis was defended at the Doctoral Evaluating Council at Academy level held at Academy of Military Science and Technology
in 8:00, on ….th August 2019
The thesis can ben found at:
- Library of the Academy of Military Science and Technology
- Vietnam National Library
Trang 3of low noised targets in passive sonars, Workshop on Maritime Communication
and Navigation (ComNavi 2016), Hanoi University of Science and Technology,
AD8302 Implemented in SONAR Interferometer, Integrated Circuits, Design,
Published, ISBN: 978-1-5386-3377-9, 5-6 Oct 2017, Hanoi, Vietnam
[5] Sang Van Doan, Jiri Vesely, Trang Cong Tran, Thanh Chi Vu, Hung Thanh Nguyen, 2017, The Measurement of Long Baseline Interferometer,
Journal of Science and Technology, Radar Monitoring Systems, Military
Technical Academy, No189, April-2018, ISSN- 1859 - 0209, pp.167-176
[6] Tran Cong Trang, Đoan Van Sang, Nguyen Thanh Hung, Tran Van Hung,
2018, Application of the MUSIC algorithm for DOA estimation in passive sonar,
Journal of Military Science and Technology No.56, (8/2018), ISSN-1859-1043,
pp.105-114
[7] Sang Van Doan, Trang Cong Tran, Van Duc Nguyen, DOA Estimation of
Underwater Acoustic Signals By Using Non-uniform Linear Arrays, 4th EAI
International Conference: INISCOM 2018, Danang Vietnam 27 August 2018, ID
339568416
Trang 5INTRODUCTION Urgency of the Thesis
Passive sonar systems play an important role in the naval warships, especially in the submarines due to the superiority of operation confidentiality Currently, advances in electronic technology and signal processing allowed to
design modern passive sonar systems with superior features
In the world, the sonar technology has developed rapidly since the beginning
of the 21st century [67], especially, in the United States and the European countries such as United Kingdom, France, Rusia [80] In Asian countries, such
as Japan and South Korea achieved success in civil sonar techniques Recently, China has invested heavily and also obtained many achievements in research and development of military sonars [23, 78] These countries have been developing many modern sonar systems, focusing on the research of passive sonar In the world and in Vietnam, the research results of passive sonar in the military application are strictly secuređ, hence it is very difficult to evaluate fully and
exactly the situation and tendency development of passive sonar technology The Objective
- Theoretical analysis and evaluation of the target positioning methods and DOA estimation methods for ship’s passive sonar systems based on mathematical
models and simulation tools
- Proposing novel solutions to increase the accuracy of DOA estimation for improvement of target location determination accuracy The proposed algorithm
is presented using the mathematical description, simulated using Matlab to validate the theory, and the experimental measurement is performed to validate and evaluate theory and simulation
The Subject and Scope
This paper studies the positioning methods, measures the angle of arrival of the sonar signal source for the object of passive sonar on ship; builds sound broadcast signal model, evaluates the ability to estimate angle of arrival by simulation It also constructs a negative antenna network configuration using MUSIC algorithm to improve the accuracy of the angle measurement of many different sonar signal sources
This paper also simulates, evaluates, compares the ability to determine the angle of incidence according to the MUSIC algorithm with algorithms CB (Conventional Beamforming) and MVDR (Minimum Variance Distortionless
Trang 6Response - meet no distortion of minimum variance), proposes solutions to improve the angle measurement accuracy to DOA to improve the target positioning accuracy for passive sonar objects on ship
Scientific and pratical significance of the Dissertation
Scientific significance
- Contributing the recommendation and conclusion for the applicability of the target positioning methods, and DOA estimation algorithms in passive sonar systems
- Establishing the signal model of the arrays ULA, UCA, and NLA for the performance of DOA estimation using CB, MVDR, MUSIC algorithms
- Proposing the Spatial Smoothing MUSIC algorithm to estimate DOAs of multiple coherent and non-coherent signal sources with higher accuracy and resolution
- Developing the system with a hydrophone NLA combined with proposed Spatial Smooth MUSIC algorithm to estimate DOAs of multiple signals emitted from various coherent and/or non-coherent sources, thereby the accuracy of target positioning determination could be increased
Pratical significance
- Performance of experiment measurement in lab water tanks to demonstrate and prove the applicability of DOA estimation using phase comparison and TDOA methods
- Performance of experiment measurement in Nha Trang Bay to demonstrate and prove the applicability of DOA estimation using MUSIC algorithm
Research methodology
Using research methods, mathematical analysis of known algorithms on target positioning and DOA estimation; using simulation tool in Matlab environment to verify theory; comparing and evaluating errors for each method Establishing the practical experimental measurements to evaluate and verify the theory and simulation; proving the applicability of the proposed solutions and methods
Structure of the Thesis
The Ph.D thesis consists of 123 pages, 82 figures, 11 tables Structure of the
dissertation includes the introduction, 3 chapters, and the conclusion Chapter 1:
Overview of passive sonar and target location methods in ship’s passive sonar
systems Chapter 2: Technical solutions for improvement of the accuracy of DOA estimation in ship’s passive sonar systems Chapter 3: Application of
MUSIC algorithm for improvement of the angle measurement accuracy in ship’s passive sonar systems
Trang 7Chapter 1 OVERVIEW OF PASSIVE SONAR AND TARGET LOCATION METHODS IN SHIP’S PASSIVE SONAR SYSTEMS
1.1 Review of associated national and international publications
From the beginning of the 21st century, projects and scientific research publications for sonar in Vietnam have just been considered more Some studies focused on the tasks of technical mastery, equipment improvement, small-scale testing Since the 2010s, the number of military sonar research has increased significantly, due to the operation urgency of equipped new submarines [12] Currently, there are many research projects on passive sonar in the world to improve detection ability and accuracy of target positioning Most studies focus
on proposing target positioning algorithms based on buoy antenna or towed antennae system In particular, the algorithm to improve the accuracy of the direction finding of the signal source is the core factor to improve the accuracy of the target location
1.2 Sound propagation characteristics
1.3 Target positioning methods in passive sonar systems
HDPR method is considered by European submarine designer and used in many passive sonar systems because it has the ability to locate targets quickly with a quite high accuracy However, the HDPR method is limited in deployment range and frequency because r must be less than /10 Modern submarines often have sonar systems arranged along the hull to determine the target distance TMA method is a classic method that has been used for a long time for submarines Although it is possible to determine the target position with only the information
of the direction But the method accuracy is relatively low In the field of passive sonar, the triangulation method is very effective, especially passive sonar systems arranged by floating buoys or near-shore observation stations or carried by surface vessels For submarines, the triangulation method is applied with an additional towed antenna system
1.4 Formulation of Problem and Research Conditions
For the purpose of studying to find solutions to improve the accuracy of angle measurement for improving the accuracy of the sonar target localization, the research problems are considererd under the following conditions:
1 The problem of target positioning of a ship's passive sonar system was studied with the assumption that the signal in acoustic noise was detected
2 Underwater acoustic signal with carrier frequency f 0 = 4 kHz was used in the Dissertation for simulation and experimental measurement because the carrier
Trang 8frequency f = 4 kHz is a medium value in frequency detection band of passive
2 The error of the target coordinate measurement of the passive sonar device depends on the two main parameters that are the angle measurement error and the time difference measurement (for the HDPR method) Therefore, in the Dissertation, the direction finding is selected to research and the solutions to improve the accuracy of measuring the angle of arrival of the sonar signal source are proposed to achieve the objectives of the Dissertation is to improve the accuracy of target location determination of the ship’s passive sonar system There are two problems of Dissertation that are needed to resolve:
1 Analyzing and evaluating the ability to DOA estimation by amplitude, phase comparison and TDOA (Time Difference Of Arrival) methods, and CB (Conventional Beamforming) algorithm, MVDR (Minimum Variance Distortionless Response) and then going deep in analyzing MUSIC by simulation, aiming to apply MUSIC algorithm to achieve the purpose of improving the angle measurement accuracy of passive sonar system This problem is solved in chapter 2
2 The studying and proposing an appropriate antenna network configuration with optimum MUSIC algorithm for improving the accuracy the angle of arrival
of the sonar signal source by simulation and empirical measurement to elucidate the research results is partially solved in chapters 2 and 3
Chapter 2 TECHNICAL SOLUTIONS FOR IMPROVEMENT OF THE ACCURACY
OF DOA ESTIMATION IN SHIP’S PASSIVE SONAR SYSTEMS
Estimating the angle of arrival is a very important step in a sonar system, especially in a passive sonar In the thesis, the frequency of 4 kHz is selected to perform the simulations, and practical measurement, because this is a frequency which is consequently used in many passive sonar systems In addition, passive sonars in our submarines and many countries operate on this frequency [6]
Trang 92.1 DOA estimation methods in passive sonar
The ability to measure the phase of a narrow-band signal has been verified by experimental measurements and has been published in [4] The measurement results show that the phase measurement method could be applied for passive sonar to measure the angle of arrival of a narrow band signal with relatively high accuracy The DOA estimation method by measuring the time difference of arrival was investigated using experiment measurement and published in [3]
2.2 Hydrophone array model
2.3 DOA estimation of underwater acoustic signals
2.3.1 DOA estimation using CB algorithm
The CB algorithm with hydrophone ULA reduces the main beamwidth and steers it to various direction electrically to received simultaneously multiple signal sources and distinguish them The power p of the output signal of the ULA is calculated as:
2.3.2 DOA estimation using MVDR algorithm
For optimization of the DOA estimation problem, Capon proposed a technical resolution to improve the DOA estimation performance [14] Instead of the
Trang 10finding maximum of response power at the output of hydrophone array, Capon used a weighted vector to make minimum power of noise power at the output, this technique so-called MVDR The MVDR condition can be written as:
1)()(
Thus, the optimal weighted vector is obtained by:
H x
x
) ( ) (
) (
1 1
a R a R w
P
) ( ) (
1 )
2.3.3 DOA estimation using MUSIC algorithm
The MUSIC algorithm was firstly presented in 1979 by R O Schmidt, then it was researched and developed in various scientific publications [58], [59], [60]
In Vietnam, the development and application of the MUSIC algorithm for DOA estimation of radar, communication, wifi, and sonar signal were still new problems Some recent publications for the direction estimation of radiated signal source by MUSIC algorithm were considerable such as [19], [20], [29], [30], [34], [38], [54], [68] The antenna array parameters were changed to improve
Trang 11performance of the DOA estimation [19], [54]; In other contributions, the MUSIC algorithm was used for the direction finding of wifi and sonar signal sources [30], [38] [20], [29], [34], [34], [68] These researches were investigated only by computer simulations Bhaumik Barodia [19], Pooja Gupta, S.P Kar [54] performed investigating on ULA with changing the number of elements, spacing distance between elements, number of snapshots, the direction of non-coherent narrow-band signals Carine ElKassis et al analyzed theoretically, simulated on NLA with missing elements, and showed that NLA had several advantages in comparison with ULA that had the same number of sensor elements [20], [29] Eva Kwizera researched direction finding of wifi signal with the frequency of 2,4 GHz, Dcom 7,5 GHz on an ULA and a NLA [30] In addition, I M Amine and
B Seddik considered wifi sources with the frequency of 2,4 GHz for UCA [38]
In 2009, Serap Cekli and Hakan Ali Cirpan used two receiving sensors to measure the direction of radiated sources, and by simulation, they demonstrated that the MUSIC algorithm was more effective than phase measurement method [68] In
2014, H Zhang et al measured angles of sonar signal sources using NLA for coherent signals The mentioned contributions above were not yet a thorough solution to accurately measure correlated signals
non-DOA estimation using MUSC algorithm was established based on computing eigenvectors and eigenvalues of the covariance matrix of the receiving signals from the sensor array [48], [50] The covariance matrix was computed by correlation function between signals from each sensor The signal model was described in detail in section 2.2, Chapter 2 of the Dissertation, which is summarized as:
R H
where, H
E ss
R s , R nEnnH is covariance matrix of coresponding signals s and
noise n Thus, signal subspace matrix Rs has size of P × P, noise subspace matrix
R n has size of M × M, and steering vector A has size of M × P Matrix A H is
conjuction matrix of the matrix A Hence, the matrix R x must have size of M × M
The matrix R n is correlated matrix between Gausian noise, so R n is describded as follow:
I
Trang 12where, I is unit matrix, 2
is covariance of noise
In order to the sensor array to distinguish all signal sources, the number of
radiated sources must be smaller than the number of sensor elements, P < M
From the determination of the subspace of the signal and noise, it is possible to
identify the steering vector a(i ) where i = 1, 2, …, P According to the
characteristics of noise, the subspace vector of noise is always orthogonal to the steering vectors at each corresponding angle 1, 2, …, P Thus, The MUSIC spectrum is given by the following equation:
) ( )
(
1 )
music
where, E n is eigenvector for Gaussian noise subspace At the angles 1, 2, …, P,
the MUSIC spectrum P MUSIC has extremely maximum values, that means the
spectrum P MUSIC has peaks corresponding to the angle of arrival of signal sources
2.4 Conclusion of Chapter 2
The results obtained in Chapter 2 could be concluded:
1 The classic methods using amplitude, phase comparison and time difference
of arrival were used to measure DOA of one signal source The methods were quite fast in processing, required several sensor elements (low cost) Thus, the DOA measurement accuracy was dependent on the main beamwidth However, the methods were difficult to measure the angles of multiple sources simultaneously
2 The CB, MVDR, and MUSIC were the modern algorithms, which are the possibility to estimate and distinguish the multiple signal sources simultaneously These methods have the potential to improve accuracy in measuring the angle of arrival of the multiple sonar signal sources
3 The CB algorithm was utilized for determining DOA of multiple signals, but it has a disadvantage that the DOA estimation accuracy was not high due to wide beamwidth
4 The MVDR algorithm was used to reduce the beamwidth, provided better angle discrimination, thus it had higher accuracy than the CB algorithm It eliminated the bias in estimating multiple sources However, the algorithm was complicated; therefore it required a large time consumption
5 The MUSIC algorithm had a higher accuracy of the DOA estimation than
CB and MVDR algorithms, but the algorithm is more complicated and must estimate the number of signal sources being monitored To apply the algorithm for the purpose of improving the accuracy of angle measurement, it was necessary
Trang 13to solve several challenges such as using the minimum number of elements, reducing the calculation volume, the DOA ambiguity problem and the ability to detect multiple correlated signal sources
The simulation and analysis results of the MUSIC algorithm based on the criteria of ambiguity, ability to distinguish signal sources, especially the analysis
of the accuracy of angle measurement with changing direction of the signal, changing the spacing distance between the elements, changing the number of sensor elements, changing the signal frequency and changing the number of snapshots were the basis for proposing solutions on sensor array configuration and optimizing MUSIC algorithm to improve the accuracy of the angle measurement
Chapter 3 APPLICATION OF MUSIC ALGORITHM FOR IMPROVEMENT OF THE DOA ESTIMATION ACCURACY IN SHIP’S PASSIVE SONAR SYSTEMS
In chapter 3, the results of simulation and experimental studies on the DOA estimation of sonar signals using the MUSIC algorithm were presented Based on the particular analysis and evaluation of the MUSIC algorithm, its advantages and advantages were compared to the CB and MVDR algorithms Thence, a novel solution of the algorithm was proposed
3.1 Comparison and evaluation of DOA estimation performance of the MUSIC algorithm versus MVDR and CB algorithms for hydrophone ULA
Figure 3.1: Comparison of response spectrum of MUSIC algorithm versus CB
and MVDR algorithms for hydrophone ULA
Trang 14Figure 3.2: Comparison of beamwidth at level -3 dB using MUSIC algorithm versus CB
and MVDR algorithm for hydrophone ULA
Figure 3.3: Dependence of PPAR values on SNR using CB, MVDR and MUSIC algorithm
for hydrophone ULA
Conclusion: The comparisons and evaluations of the hydrophone ULA
indicated that the MUSIC algorithm provided good quality, which achieved higher accuracy and resolution than CB and MVDR algorithms
Comparison of DOA estimation ability on Sonar display
Figure 3.4: Simulation of tracking angle on Sonar display for hydrophone ULA using CB,
MVDR and MUSIC algorithms with SNR =15 dB