Nowadays, many researchers pay much attention to Multilateration localization process in air traffic control. This thesis deals with system geometry effect on target localization process. Modifications are proposed in both the sensors deployment network and the algorithm sequence used.
The work will be on a 3 steps: First, working on a small cell network taking minimum number of sensors. Different optimal sensor deployment methods will be examined and a pilot area will be chosen from a Multilateration network at Cairo International Airport, 05L Runway (RW). Simulation for a single aircraft is performed and the results are compared. Second, the deployment method that gives the best results in a small cell network will be considered and a new sensors deployment covering an area larger than the existing one is proposed, trying to have the same performance of the existing system with the advantage of decreasing the number of sensors. Finally, the proposed Multilateration network is divided into 9 main clusters and 4 backup clusters instead of using all sensors in localization process, and the cluster that has the least PDOP (position dilution of precision) is selected. To enhance the localization process only the cluster that has least PDOP < 21 is selected. Otherwise, are rejected. Space capacity and safety of Airspace Surveillance Systems are regularly increasing to meet the demands of air traffic control.
A 2-D Multilateration algorithm is proposed to accurately identify the aircraft position. It is based on the classical Two Ray propagation model. Important parameters that affect the final aircraft estimation position are presented like the path gain factor (interference factor). The proposed algorithm uses the geographic coordinates (Latitude and Longitude) which are considered more practically used in navigation than Cartesian coordinates that are used in previous algorithms in literature. In order to simulate the algorithm, the Multilateration network at Cairo international Airport is considered to be a pilot area and the results are presented.
TABLE OF CONTENTS
Abstract
Acknowledgements
Notation
List of Abbreviations
List of Symbols
Table of Contents
List of Figures
List of Tables
1. Introduction
1.1 Overview
1.2 Motivation
1.3 Thesis Contributions
1.4 Thesis Organization
2. Review Of Multilateration Air Traffic Surveillance System And Localization Problem
2.1 Introduction
2.1.1 Classifications of Localization algorithms
2.1.2 Classifications of Localization Processes
2.2 A Multilateration System Background
2.3 Localization Problem in Multilateration
2.4 Related Work
3. Geometry Effect On A Multilateration Air Traffic Surveillance System Performance
3.1 Introduction
3.2 Optimal Sensors deployment methods used for comparison
3.3 Proposed modification in a MLAT network
3.4 Proposed modification in a MLAT algorithm
3.4.1 First Modification
3.4.2 Second Modification
3.5 Simulated Scenarios
3.6 Conclusions
4. A 2D Multilateration Algorithm Used For Air Traffic Localization And Tracking
4.1 Hypotheses and Proposed Algorithm
4.1.1. Classical Two Ray Propagation Model
4.1.2. Proposed Algorithm
4.1.3. Kalman Filter Estimator
4.2 Simulation Results
4.3 Conclusions
5. Multiple Aircrafts Tracking In Clutter For Multilateration Air Traffic Surveillance System
5.1. Introduction
5.2. Proposed Algorithm for Single Aircraft Tracking in Clutter
5.3. Proposed Algorithm for Multiple Aircrafts Tracking in Clutter
5.3.1. A Proposed algorithm modification for the proposed MLAT
51 network
5.4. Simulated Results
5.5. Conclusions
6. Conclusions And Recomendations For Future Work
6.1. Conclusions
6.2. Recomendations For Future Work
References
List of Publications
English Abstract
Nowadays, many researchers pay much attention to Multilateration localization process in air traffic control. This thesis deals with system geometry effect on target localization process. Modifications are proposed in both the sensors deployment network and the algorithm sequence used. The work will be on a 3 steps: First, working on a small cell network taking minimum number of sensors. Different optimal sensor deployment methods will be examined and a pilot area will be chosen from a Multilateration network at Cairo International Airport, 05L Runway (RW). Simulation for a single aircraft is performed and the results are compared. Second, the deployment method that gives the best results in a small cell network will be considered and a new sensors deployment covering an area larger than the existing one is proposed, trying to have the same performance of the existing system with the advantage of decreasing the number of sensors. Simulation is performed again for both the total proposed network and the total existing Multilateration deployment network and the results are compared. Finally, the proposed Multilateration network is divided into 9 main clusters and 4 backup clusters instead of using all sensors in localization process, and the cluster that has the least PDOP (position dilution of precision) is selected. To enhance the localization process only the cluster that has least PDOP < 21 is selected. Otherwise, are rejected.
Space capacity and safety of Airspace Surveillance Systems are regularly increasing to meet the demands of air traffic control. Researchers are interested in many directions like optimal sensors deployment, localization and tracking algorithms. Thus, a 2-D Multilateration algorithm is proposed to accurately identify the aircraft position. It is based on the classical Two Ray propagation model. Important parameters that affect the final aircraft estimation position are presented like the path gain factor (interference factor) which is a function of the divergence factor, reflection coefficient and the path difference between the direct and ground reflected rays. The proposed algorithm uses the geographic coordinates (Latitude and Longitude) which are considered more practically used in navigation than Cartesian coordinates that are used in previous algorithms in literature. Hence, General expressions for both the latitude and longitude will be deduced. In order to simulate the algorithm, the Multilateration network at Cairo International Airport is considered to be a pilot area and the results are presented. The results of the proposed algorithm are applied to Kalman filter to achieve aircraft continuous tracking. Finally, the path gain factor effect on tracking capability is discussed.
In this thesis a Non-Bayesian single (multiple) aircrafts tracking algorithm is proposed taking into account the existence of clutter. The proposed algorithm deals with the geographic coordinates (latitude and longitude). First, a proposed single aircraft tracking algorithm in clutter is discussed. It is based on Kalman filter with a modification in the predicted state. The proposed predicted state is presented which will be a function of the state in the last three iterations; Followed by a proposed validation gate and data association analysis. The used Validation gate threshold will be variant not constant. Then a proposed multi-aircrafts tracking in clutter model is analyzed. Finally, a Multilateration network at Cairo International Airport is considered a pilot area to simulate both scenarios.
Multilateration; Air traffic surveillance systems; Air traffic Key Words control; dilution of precision; Data association and tracking algorithm; kalman filter.
ABSTRACT
Nowadays, many researchers pay much attention to Multilateration localization process in air traffic control. This thesis deals with system geometry effect on target localization process. Modifications are proposed in both the sensors deployment network and the algorithm sequence used. The work will be on a 3 steps: First, working on a small cell network taking minimum number of sensors. Different optimal sensor deployment methods will be examined and a pilot area will be chosen from a Multilateration network at Cairo International Airport, 05L Runway (RW). Simulation for a single aircraft is performed and the results are compared. Second, the deployment method that gives the best results in a small cell network will be considered and a new sensors deployment covering an area larger than the existing one is proposed, trying to have the same performance of the existing system with the advantage of decreasing the number of sensors. Simulation is performed again for both the total proposed network and the total existing Multilateration deployment network and the results are compared. Finally, the proposed Multilateration network is divided into 9 main clusters and 4 backup clusters instead of using all sensors in localization process, and the cluster that has the least PDOP (position dilution of precision) is selected. To enhance the localization process only the cluster that has least PDOP < 21 is selected. Otherwise, are rejected.
Space capacity and safety of Airspace Surveillance Systems are regularly increasing to meet the demands of air traffic control. Researchers are interested in many directions like optimal sensors deployment, localization and tracking algorithms. Thus, a 2-D Multilateration algorithm is proposed to accurately identify the aircraft position. It is based on the classical Two Ray propagation model. Important parameters that affect the final aircraft estimation position are presented like the path gain factor (interference factor) which is a function of the divergence factor, reflection coefficient and the path difference between the direct and ground reflected rays. The proposed algorithm uses the geographic coordinates (Latitude and Longitude) which are considered more practically used in navigation than Cartesian coordinates that are used in previous algorithms in literature. Hence, General expressions for both the latitude and longitude will be deduced. In order to simulate the algorithm, the Multilateration network at Cairo International Airport is considered to be a pilot area and the results are presented. The results of the proposed algorithm are applied to Kalman filter to achieve aircraft continuous tracking. Finally, the path gain factor effect on tracking capability is discussed.
In this thesis a Non-Bayesian single (multiple) aircrafts tracking algorithm is proposed taking into account the existence of clutter. The proposed algorithm uses the geographic coordinates (latitude and longitude). First, a proposed single aircraft tracking algorithm in clutter is discussed. It is based on Kalman filter with a modification in the predicted state. The proposed predicted state is presented which will be a function of the state in the last three iterations; Followed by a proposed validation gate and data association analysis. The used Validation gate threshold will be variant not constant. Then a proposed multi-aircrafts tracking in clutter model is analyzed. Finally, a Multilateration network at Cairo International Airport is considered a pilot area to simulate both scenarios
KEY WORDS
Multilateration, Air traffic surveillance systems, Air traffic control. Dilution of Precision, Data association and tracking algorithm, kalman filter.
ACKNOWLEDGEMENTS
First and foremost, I thank Allah, for all his blessings and for giving me this opportunity to carry on this work, helping me and giving me the strength and faith to accomplish.
Hence, I would like to express my sincere gratitude and appreciation to my supervisors, whose guidance and support allowed the accomplishment of this research. I really have been fortunate to be able to benefit from their experience and remarkable influence during my first steps and through the whole work.
I owe my deepest recognition and sincere gratitude to Prof. Dr. Atef Ghuniem for his invaluable guidance, precious advice, detailed and constructive comments, his generous, time and effort throughout all the stages of conducting this thesis.
Special and deepest thanks goes to Prof. Dr. Abdel-Hamid Gaafar, who helped me to complete this thesis as well as he was always there to advise, meet and talk about my ideas, he brought out the good ideas in me, thanks for his extensive discussions and continuous encouragement.
I am thankful to Asst. Prof. Dr. Hossam El-Din Abou Bakr, whose warm encouragement, I will never forget.
Last, but not least, I deeply thanks my Parents my wife and my daughters (Ratel and Icel) for giving me good life, for listening and supporting, had confidence in me when I doubted myself, without their encouragement and constant guidance I could not have finished this thesis.
Abbildung in dieser Leseprobe nicht enthalten
LIST OF ABBREVIATIONS
Abbildung in dieser Leseprobe nicht enthalten
LIST OF SYMBOLS
Abbildung in dieser Leseprobe nicht enthalten
LIST OF FIGURES
Fig.1.1: Cost benefits of MLAT
Fig.1.2: One of the main advantages of the MLAT system on SSR system
Fig.2.1: Classification of localization schemes
Fig.2.2: Triangulation
Fig.2.3: Trilateration
Fig.2.4: Multilateration
Fig.2.5: General scheme for a MLAT system
Fig.2.6: hyperbolas of 4 stations and a target located in arbitrary position. The reference station is number 1; the black one
Fig.2.7: Localization problem in multilateration
Fig.3.1. Method I, a unit square of sensors placement
Fig.3.2: Method II, Acollection of six sensors is shown in an optimal configuration at equal angle increments
Fig.3.3. Method III, Optimum geometry of a 4 sensor array
Fig.3.4: A Pilot area of Multilateration network at 05Left Runway, Cairo International Airport. (a) Using UTM Coordinator. (b) Using Google Earth Pro. (7.1), 2016
Fig.3.5: A Multilateration localization problem using a reference point
Fig.3.6: The existing MLAT total network deployed in Cairo International Airport
Fig.3.7: Proposed network of MLAT receivers for Cairo International Airport divided into clusters, (Red lines are cluster number 1, 2 and 3), (Blue lines are clusters number 4, 5and 6), (Yellow lines are clusters number 7, 8 and 9)
Fig.3.8: Flowchart of Proposed modification in MLAT algorithm…. 27 Fig.3.9: Proposed Modification flowchart in A MLAT algorithm based on PDOP
Fig.3.10: DOP coefficients of the existing, and proposed sensor network deployment for the straight line motion on the runway before takeoff
Fig.3.11: DOP coefficients of the existing, and proposed sensor network deployment for the straight line motion on 1500m high
Fig.3.12: DOP coefficients of the existing, and proposed sensor network deployment for the straight line motion on 8000m high
Fig.3.13: DOP coefficients of the existing, and proposed sensor network deployment for the circular motion on 4250m high
Fig.3.14: DOP coefficients of proposed Cluster and total sensor network deployment for the straight line motion on 8000m high
Fig.3.15: DOP coefficients of proposed Main and Backup cluster sensor network deployment for the straight line motion on the runway before takeoff
Fig.3.16: The number of clusters that have (PDOPMax<21) versus altitude in normal operation
Fig.4.1: Earth models. (a) Two Ray Spherical Earth Model. (b) Two Ray Flat Earth Model
Fig.4.2: A proposed 2D MLAT algorithm flowchart
Fig.4.3: A Pilot Area of 32 sensors MLAT network at Cairo International Airport using UTM coordinator
Fig.4.4: A complete picture of Kalman filter operation
Fig.4.5. Simulated results of an aircraft route using the proposed MLAT 2-D algorithm, KF versus the actual aircraft positions
Fig.4.6: Aircraft is disappeared and lost track for 2 iterations due to path gain factor effect before starting a new track
Fig.5.1. A single aircraft tracking in clutter, first proposed procedure
Fig.5.2: A single aircraft tracking in clutter, second proposed procedure
Fig.5.3: Several measurements zi in the validation region of a single target. The validation region is an ellipse centered at the predicted state. Any of the shown measurements could have originated from the target (or none if the target is not detected)
Fig.5.4: Two validation gates are intersected and having the same minimum distance measurement (O4) for both targets
Fig.5.5: Proposed Multi-Aircraft tracking algorithm block diagram
Fig.5.6: Proposed Multi-Aircraft tracking algorithm block diagram for the proposed MLAT network
Fig.5.7: Data association process for a single aircraft tracking in clutter within the pilot area
Fig.5.8: Data association processes for two aircrafts are tracked in clutter within the pilot area
LIST OF TABLES
Table 3.1: DOP Ratings
Table 3.2: Sensor Density Comparison of The Total Existing and Proposed Network
Chapter 1 Introduction
This chapter gives a brief overview of the work. It includes the context in which the thesis was developed and the main motivations. At the end of the chapter, the work structure for the thesis is presented.
1.1 Overview
Nowadays, travelling by airplane can be considered safer than by car or train, because there are very strict requirements in terms of safety. One of them is the surveillance systems that are required to monitor every target in the airspace. These surveillance systems must guarantee that the large number of airplanes travel simultaneously in the air safely. As the amount of passengers increase, Global air traffic also increases which demands doing an effort to do the following:
- Increase efficiency
- Streamline operations
- Minimize infrastructure cost
- Improve safety
To meet these safety requirements, the airspace surveillance technology is developed to improve the capacity with the performances of new telecommunication systems. So, it is needed to turn away from the traditional radars and looking towards a different technology. Every Air Navigation Service Provider (ANSP) needs to upgrade his systems to provide the surveillance service with the higher level of safety to all aircrafts in the airspace. It must be performed in an efficient way, and with limited funds, so every new system implementation and upgrades must be carefully studied and analyzed by a cost benefit analysis.
There are three categories for surveillance defined by Euro-control 1:
- The first is an independent and non-cooperative surveillance system used to track all aircrafts. This can be provided by the Primary Surveillance Radar (PSR) system, which is the oldest surveillance system. This system is not the most efficient one, but it is still recommended to be kept, because it is the only way of detecting an aircraft if the aircraft's transponder fails.
- The second is an independent cooperative surveillance system used to track cooperative aircrafts, which means that even though it is required that the aircraft sends a reply signal; the localization is calculated in the ground station. There are two systems that comply with this principle. The first system to be used in this category was the secondary surveillance radar (SSR), but more recently a new system which named Multilateration (MLAT) is appeared and can replace the SSR.
- The third is a Dependent cooperative surveillance, which means that the localization information is supplied by the flying aircraft, instead of being calculated by the ground station. The system that supports this principle is automatic dependent surveillance (ADS), which allows the ground station to receive a message with the aircrafts location measured by their equipment.
The focus of this thesis is in the independent cooperative surveillance category, more specifically in the latest innovation which is MLAT. The MLAT systems nowadays are an important part of air traffic control technological infrastructure. They are used by air traffic controllers to know about the position and identity of all aircrafts and equipped vehicles in the coverage area of interest.
The MLAT system can replace the SSR in the independent cooperative surveillance system category, because it improves the efficiency, accuracy, infrastructure costs and safety. Another main advantage of the MLAT system is the possibility to monitor the airplanes on the grounds at an airport. Concerning MLAT and SSR costs, a study was performed by ERA 2 (ERA is a surveillance system's manufacturer). The results provided by the study are shown in Fig. 1.1. In terms of costs, there is a large difference in the acquisition and maintenance price, which will decrease a lot the expenses for the ANSPs. It shows that in terms of costs, the MLAT solution is better, being one of the reasons that it will be used in the long term.
Abbildung in dieser Leseprobe nicht enthalten
Fig. 1.1: Cost benefits of MLAT (extracted from 2 ).
Beside the costs, there is another main advantage which also gives preference to MLAT. Safety is increased because the MLAT system by itself is redundant, and even if some parts of the system fail it will continue to work, unlike SSR that if the radar itself needs to be maintained or fails, the system will stop working as shown in Fig. 1.2.
In Egypt, there is already an operational MLAT system at Cairo International airport, which is used in this thesis to give more details to the architecture and even to collect data to improve the analysis.
Abbildung in dieser Leseprobe nicht enthalten
Multilateration Multi late ration
Fig. 1.2: One of the main advantages of the MLAT system over SSR system 3.
1.2 Motivation
Every ANSP buy the MLAT systems from a manufacturer and, even though manufactures have the responsibility that the system meets the requirements, it is also important that the ANSP gets some knowhow of the implemented systems. So it is necessary to judge how accurate the system performance. The most important requirements from the MLAT surveillance system are aircraft safety and tracking accuracy.
The existing MLAT system at Cairo International Airport faces a big problem. If any two sensors have failure at the same time, the MLAT system become out of service and makes shutdown which is not practical in navigation usage. Also the final position estimation is affected by many factors like geometry effect and path gain factor (interference factor); which can affect the tracking process. The present work will focus on these problems trying to get reasonable solutions.
1.3 Thesis Contributions
The present research attempts to overcome the problem of emergency shutdown of the existing Multilateration air traffic surveillance system at Cairo International airport due the failure of any two sensors at the same time. A new network deployment is proposed to mitigate the influence of geometry on the final estate estimation. Then it is proposed to divide the total network into 9 clusters and any cluster has a failed sensor will be isolated while the MLAT system is still running. After that the MLAT algorithm will run in parallel manner on all clusters and the cluster that has least position dilution of precision (PDOP) will be selected to be the most trusted estimated position with minimum error. To enhance the localization process, a threshold is proposed for PDOP. If less the estimated position will be screened out to the air traffic controller, else it will be rejected and return back to the initial state to start sensing for a new target.
Also the research tries to examine and determine another factor which affects the localization process like path gain factor. And a 2D MLAT algorithm is proposed which uses geographic coordinates (latitude and longitude). General expressions for the both the latitude and longitude are deduced. Simulations are done taking into account the path gain factor effect on the aircraft tracking process. Finally, the research attempts to discuss single and multiple aircrafts tracking processes using Multilateration air traffic surveillance system. A new algorithm is proposed to track both scenarios taking into account the existence of clutter or false alarm measurements.
1.4 Thesis Organization
The outline of the rest of this thesis is as follows:
- Chapter 2: Review of Multilateration Air Traffic Surveillance System and Localization Problem
This chapter presents briefly an overview on the classification of both localization processes and algorithms. Then the multilateration system architecture and its operational concept will be disscussed. After that the localization problem in multilateration will be presented. Finally, the literature of relevant algorithms and hypothesis proposed earlier in multilateration will be displayed.
- Chapter 3: Geometry Effect on a Multilateration Air Traffic Surveillance System Performance
This chapter discusses the influence of network deployment geometry on the final estimated position accuracy for different scenarios. A new deployment for the multilateration network will be proposed and two proposed modifications in the used algorithm will be presented.
- Chapter 4: A 2D Multilateration Algorithm used for Air Traffic Localization and Tracking
In this chapter 2-D localization and tracking algorithm based on the classical two ray propagation model will be proposed. General expressions for both the latitude and longitude are deduced. Finally, tracking scenario is discussed in case of taking the path gain factor into account.
- Chapter 5: Multiple Aircrafts Tracking In Clutter For Multilateration Air Traffic Surveillance System
In this chapter a non-Bayesian algorithm is proposed for both single and multiple aircrafts tracking incase of clutter existed. The proposed algorithm is based on Kalman filter approach with a modification in the predicted state.
- Chapter 6: Conclusions and Recommendations
It discusses the conclusions of the thesis and recommendations for the future work
Chapter 2
Review of Multilateration Air Traffic Surveillance System and Localization Problem
In this chapter, the classification of localization schemes will be discussed generally in order to understand the concept of multilateration. Then the multilateration system description and the problem of localization process are discussed. Finally, the relevant algorithms proposed earlier are presented.
2.1 Introduction
The existing localization schemes are classified into Localization algorithms and Localization Processes as shown in Fig. 2.1. The localization algorithms [4-17] are classified into three major categories as follows:
- Centralized and distributed approaches.
- Range free and range based approaches.
- Anchor free and anchor based.
While the localization processes 4 are also classified into three major categories
Abbildung in dieser Leseprobe nicht enthalten
b) Classification of localization Processes.
Fig.2.1: Classification of localization schemes.
which are as follows:
- Triangulation
- Trilateration
- Multilateration
2.1.1 Classifications of Localization Algorithms
The three categories of localization algorithms are discussed briefly in this section.
A. Centralized and Distributed Approaches
In centralized approaches all computations are done in the central processing unit, so all nodes have to communicate to base station. In distributed approaches, computations are distributed among sensor nodes. It is clear that communication process consumes more energy in centralized approaches as compared to communication cost in distributed approaches.
B. Range free and range based approaches
In range free algorithms only use connectivity information to determine node's location. In range based most important used techniques are Time Difference of Arrival (TDOA), Time of Arrival (TOA), Angle of Arrival (AOA) and Received Signal Strength Indicator (RSSI).
C. Anchor free and Anchor based approaches
Anchor nodes are defined as those nodes that know their coordinates by GPS or manual placement so; we can get the global coordinates. But GPS receivers are expensive and can't be used indoors.
2.1.2 Classifications of Localization Processes
The three categories of localization processes are discussed briefly in this section.
A. Triangulation
Triangulation 4 is used to determine the position of target node; the process is done by measuring the angle of the target node to the anchor node as shown in Fig. 2.2. On the other hand triangulation can be the angular distance between the three anchor nodes forming a triangle and the target node inside triangle.
Abbildung in dieser Leseprobe nicht enthalten
Fig 2.2: Triangulation 4
B. Trilateration
Trilateration 4 is the process of finding the position of a target node by measuring the distance from at least three anchors (i.e. the intersection point of three circles gives the position of the target node) as shown in Fig. 2.3.
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.3: Trilateration 4
C. Multilateration
In multilateration 4, more than three anchor nodes are used to find the position of the target node as shown in Fig. 2.4. Actually, most of authors refer to trilateration as a subset of multilateration process with (N =3 sensors).
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.4: Multilateration 4
2.2 A Multilateration System Background
A multilateration (MLAT) system 18 is a cooperative/independent positioning system.1 It is composed of a set of synchronized receiving stations (stationary) and one or more interrogating stations deployed throughout a coverage area connected through a communication network to a computing central processing unit where a set of algorithms are executed to obtain the position of either surface or air targets as shown in Fig. 2.5.
The most used standard version of MLAT systems measure TOA (time of arrival) of signals which are transmitted from the aircraft and received by a set of ground received stations that are deployed around the coverage area. A minimum number of 3 sensors are needed to have a 2D position while a minimum number of 4 sensors are required to have a 3D position. The TOA measurements are sent to CPS (Central Processing Subsystem) to be processed and locate aircrafts or vehicles position.
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.5: General scheme for a MLAT system 18
The received signal is detected, decoded and processed to carry out the identification of aircraft. MLAT operates based on queries and the target replies with a coded signal. FDM (Frequency Division Multiplexing) is used to separate the interrogations of ground station to aircraft 1030 MHz from the aircraft replies to ground station (1090 MHz). The reply signals can be in a mode A, C or S 1. For noiseless case TOA measurements are defined as follows:
Abbildung in dieser Leseprobe nicht enthalten
Where, te is the time of signal emission, c is the velocity of light, θp=[x, y, z]T is the unknown aircraft position , vi = [xi, yi, zi]T is the known position of ith receiver and [.]T means vector transpose. But te is unknown, so it is prefered to use TDOA (Time Difference of Arrival) technique by taking sensor number 1 to be a reference as follows:
Abbildung in dieser Leseprobe nicht enthalten
Theoretically, for a MLAT system composed of N stations, it is possible to obtain a total of hyperbolic equations in the form of (2).
Abbildung in dieser Leseprobe nicht enthalten
The solution of this hyperbolic system of equations is the spatial point of intersection of all this hyperboloids as shown in Fig. 2.6.
Hence, MLAT system collects a certain set of measurements from a set of received stations then obtain a target position by solving a numerical problem (parabolic equations), so this process can be seen as a domain change in which measurements (a domain) are converted into another domain (spatial). But actually in MLAT, researchers are interested in the localization process, its accuracy and its reliability. They found that system accuracy depends on 3 factors which are:
- System geometry.
- Measurement noise.
- Robustness of the localization algorithm.
Abbildung in dieser Leseprobe nicht enthalten
Fig.2.6: hyperbolas of 4 stations and a target located in arbitrary position 18. The reference station is number 1; the black one.
The final MLAT system accuracy (called the operational system accuracy) can reach the theoretical one depending on its efficiency. The theoretical efficiency is defined as the best accuracy that can be achieved by MLAT system composed of a set of stations drawing a specific geometry where each station measure some parameters with a statistically definable measurements accuracy. The theoretical system accuracy can be provided by:
- The measurement accuracy.
- DOP (Dilution of Precision).
Dilution of precision (DOP) is a unitless deterministic quantity that represent the quality of the system geometry for calculating the target position in a particular point in the coverage area or a volume.
The localization algorithms can be summarized into 2 families:
- The open form algorithms (also called iterative algorithms).
- The closed form algorithms (also called direct algorithms).
The open form algorithms are statistically efficient (unbiased and minimum variance) but not numerically efficient (unstable convergence). While the closed form ones are numerically efficient (stable convergence) but not statistically efficient (biased and highly dispersed with respect to CRLB). So, it is clear that there is no superior algorithm that provides under all conditions a solution that is both statistically and numerically efficient. Some authors [19, 20] proposed a combined algorithm of closed and open form algorithms. They used the closed form algorithm to get the starting point to the open form algorithm.
2.3 Localization problem in Multilateration
The aircraft/vehicle on board transponder device emits, at time te a reply signal which is received by N ground received stations at different time instants. Each station measures the time of arrival and sends it to the CPS where the localization algorithm is run to calculate the target position. The received signal contains some amount of noise nprop because of the propagation from the target transponder to each sensor as shown in Fig. 2.7.
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.7: Localization problem in multilateration 18
The TOA measurements are affected by a certain amount of noise [21-23] which can be summarized into the propagation noise, the receiver instrumental errors, the quantization effects in the receiver unit, and the synchronization errors. These measurements can be expressed as follows:
Abbildung in dieser Leseprobe nicht enthalten
Where, TÔA is the estimated/measured TOA quantity, ri(9p) = I9p - v jl , is the transponder range to ith RX station and 8terri is the TOA error term in ith RX station. Then, the noisy TDOA measurement can be expressed as follows
Abbildung in dieser Leseprobe nicht enthalten
Assuming that the noise term, se = [5terr2,1 , 5terr3,1 , …, 5terrN,1 ] is a random variable ~ Gaussian distribution with zero mean and variance JsT.
Abbildung in dieser Leseprobe nicht enthalten
2.4 Related Work
Multilateration process is a rich topic so many researchers are interested in it. A survey is done and found that there are two common points most of researchers are interested in them. The first one is how to design and deploy an optimum multilateration network to have minimum error in the estimated position. The second point is developing new algorithms to be more accurate in the estimating position and more immune to noise.
Authors in 24 studied and developed some strategies to design and deploy Mode-S multilateration systems. Authors in 25 are interested in the optimal placement of M planar sensors to have the best expected source location estimate. Authors in [26, 27] found that an equiangular surrounding of the target by an arbitrary number of sensors is an optimal sensor placement.
Authors in 28 derived properties of the Cramer-Rao Lower Bound (CRLB)2 and designed optimum sensor arrays that minimize CRLB for 2D and 3D localization from information got from TDOA with the assumption of white measurement noise with uniform covariance. Authors in 29 proposed an adaptive geometric optimization technique that removes the singularities with a model of software-based rotational geometry. Authors in 30 optimized the sensor placement for mobile sensor networks by proposing a motion coordination algorithm that directs the mobile sensor network to an optimum sensor placement.
Authors in 31 proposed an open form algorithm called regularized location estimator (RLE) for TDOA (time difference of arrival) MLAT system. This algorithm is based on Maximum likelihood estimation. They found (using a desktop computer with 4 GB RAM and a 2.1 GHz processor) that the average time for estimating a position was only 8 micro seconds.
Authors in 32 used a regularized method to solve localization problem in MLAT and found that for λ = 0.1 (regularization parameter value) it was sufficient to find satisfactory results and if the amount of ill-2 CRLB expresses a lower bound on the variance of estimator's parameter. conditioning become small, λ should be smaller. Authors in 33 tried to improve location estimation of RFID (Radio Frequency Identification) passive tag in a NLOS (Non-Line of Sight) environment using MLAT, KF (kalman filter) and EKF (Extended kalman filter) technologies and found that EKF has lower deviation as compared to KF. Authors in 34, used a synchronous and spatially separated pairs of emitters and one sensor trying to minimize the difference between signal based TDOA measurements from system and estimated TDOA measurements made by calculations based on a given sensor positions by means of Least squares optimization.
Authors in 35, proposed a two closed form localization algorithms for MLAT system to make a quick rejection of false time measurements based on coding theory. Authors in 36, tried to maximize the coverage for a given accuracy using genetic algorithm but unfortunately they forgot to take into account that the localization is also depends on the construction of the site.
Authors in 37 proposed a Bayesian data association method based on the particle filter idea and the joint probabilistic data association (JPDA) hypothesis calculations. In 38 a number of the problem formulations are reviewed, including two-dimensional asymmetric single and multi-assignment problems. Authors in 39 proposed a computationally efficient algorithm based on joint probabilistic data association for target-measurement correlation.
In 40 the joint probabilistic data association (JPDA) technique is revisited and proposed, moreover, a novel solution based on recent developments in finding the m-best solutions to an integer linear program is proposed. Authors in 41 compared the results of the algorithm Global Nearest Neighbor (GNN) with Suboptimal Nearest Neighbor (SNN) that are used for Multiple Target Tracking (MTT). In 42 efficient data association algorithms are proposed for multi-target tracking.
Chapter 3
Geometry Effect on a Multilateration Air Traffic Surveillance System Performance
In this chapter, the proposed modifications of a MLAT system are performed on both the sensors deployment network and the algorithm sequence used. First, Applying a Multilateration algorithm to different optimal deployment methods with minimum number of sensors (4 sensors) and results are compared to a pilot area simulated results. The pilot area is chosen from the existing multilateration network at Cairo International Airport 05L runway (runway direction). Second, the method that will give the best results in a small cell network simulation will be considered in the proposed total network deployment design. Trying to achieve the same performance of the existing deployment with the advantage of decreasing the number of sensors used (24 sensors instead of 32 sensors). Third, to overcome the emergency multilateration system shutdown problem due to failure of any 2 sensors in the network at the same time, the total network is proposed to be divided into 9 main clusters and 4 backup clusters instead of using all sensors in localization process, and the cluster that has the least PDOP (position dilution of precision) is selected. Finally, it is proposed to limit the error to 21 and only the cluster that has least PDOP < 21 is selected. Otherwise, are rejected.
3.1 Introduction
System geometry 18 refers to the spatial arrangement of receiving stations (Rx) relative to the coverage area. The system geometry influence on the system accuracy is directly related with both eccentricity and perpendicularity of the hyperbolas derived from each TDOA (Time Difference of Arrival) measurements. Every TDOA measurement contains some amount of error that obeys a Gaussian distribution of zero mean and certain standard deviation (σ). This fact makes TDOA measurements to be non-exact and can be lying on a strip centered at the theoretical curve in case of 2D solution. The strip width depends on the standard deviation magnitude of each measurement error distribution. Hence, the target can be located with a given probability in an area called "Uncertainty area" in case of 2D solution and in "Uncertainty volume" in case of 3D solution. The size of this area is different due to eccentricity and perpendicularity of the hyperbolas. Both eccentricity and perpendicularity of hyperbolas are mathematically comprised in a DOP factor that defines the quality of the system geometry (i.e. DOP defines how much the system geometry magnifies the measurement errors in the system accuracy).
In section 3.2, a set of optimal sensor deployment methods that will be used in comparison will be presented. A new network deployment design is proposed in section 3.3. In section 3.4, a modification in multilateration algorithm to use clusters and use PDOP coefficient in localization process will be proposed to enhance system performance.
3.2 Optimal Sensor Deployment Methods used for Comparison
A survey has been made to choose the best 3 optimal sensor deployment methods. In the following paragraphs a summary of a small cell sensor deployment methods used in comparison and the pilot area that will be under study will be presented.
In method I, Authors [26, 27] said that the optimal geometry is achieved when the target is located at the center of a unit square of sensors placement as shown in Fig. 3.1. It is reasonable to assume that the good localization performance can be achieved when the target is located anywhere inside the unit square and they proved that this sensor configuration is optimal which minimizes CRLB.
In method II, Authors 25 show that for sensors deployment to be optimum, we need two conditions to be satisfied. First, the sensors should be placed as far from the expected source position as possible. Second, the sensors should be arranged in a splay configuration in which neighboring sensors are separated by equal angle increments as shown in Fig. 3.2.
Abbildung in dieser Leseprobe nicht enthalten
Fig.3.1. Method I, a unit square of sensors placement.
Abbildung in dieser Leseprobe nicht enthalten
Fig.3.2. Method II, A collection of six sensors is shown in an optimal configuration at equal angle increments.
In method III, Author 43 derived an optimum 3D sensor placement. The reference sensor is assumed to be located at point g1 = [0, 1, 0]T, where [.]T means vector transpose and the remaining 3 effective sensors should be symmetric to both the source and the reference sensor as well as among themselves; where they are equally spaced on the thick circle as shown in Fig. 3.3. They derived after some straight forward calculations that, For a CRLB – optimum; α = 98.9° (α is the angle between y- axis and sensor located at point g2). Let point g2 be in xy-plane. The position vectors gi (1< i < 4) are as follows:
Abbildung in dieser Leseprobe nicht enthalten
Fig.3.3. Method III, Optimum geometry of a 4 sensor array 43.
[...]
1 Mode A is used for aircraft identification. Mode C is used for aircraft altitude. And mode S is used to make selective interrogations.
2 CRLB expresses a lower bound on the variance of estimator's parameter.
-
-
-
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X.