Integrated Display and Simulation for Automatic Dependent Surveillance-Broadcast and Traffic Collision Avoidance System Data Fusion

Yanran Wang, Gang Xiao, Zhouyun Dai

TL;DR

This paper presents an integrated system that fuses data from Automatic Dependent Surveillance-Broadcast (ADS-B) and Traffic Collision Avoidance System (TCAS) to improve airspace surveillance and collision avoidance. By developing a minimum ADS-B system with a novel Variational Bayesian-Interacting Multiple Model (VB-IMM) algorithm, the researchers demonstrate enhanced accuracy and reduced false alarms compared to individual systems, paving the way for safer air traffic management.

What Problem Does This Paper Address?

The increasing complexity of air traffic necessitates more robust surveillance and collision avoidance systems. While ADS-B provides valuable surveillance data and TCAS offers direct collision avoidance, their independent operation has limitations. Research on fusing these two systems is relatively scarce, leaving a gap in understanding how to leverage their combined strengths. Specifically, the paper addresses the need for improved prediction accuracy in TCAS, a reduction in false and missed alarms, and a more resilient system that can function even if one data source is partially unavailable. This is crucial for the development of next-generation air traffic systems.

Methods

The core of this work lies in the development of an "ADS-B minimum system" with two main components: ADS-B In (receiving and processing data) and ADS-B Out (transmitting data).

For ADS-B In, the key innovation is a data fusion model employing a Variable Sampling Variational Bayesian-Interacting Multiple Model (VSVB-IMM) algorithm. This algorithm combines:

  • Variational Bayesian (VB) methods: Used to estimate time-varying noise variances, which is crucial for adapting to changing environmental conditions and improving filter performance.
  • Interacting Multiple Model (IMM) algorithm: A standard technique for tracking targets with uncertain motion by maintaining multiple models (e.g., Constant Velocity, Constant Acceleration) and combining their estimates.
  • Variable Sampling: The algorithm adapts its sampling rate based on the estimated position variance and Navigation Accuracy Category (NACp) of the incoming ADS-B data, allowing for more intensive tracking when accuracy is lower or the intruder is closer.

The system also incorporates:

  • Coordinate Transformation: Converting data from different coordinate systems (WGS-84, ECEF) into a unified frame.
  • Time Synchronization: Aligning data from ADS-B and TCAS, which may have slightly different update rates or delays.
  • Optimal Information Fusion Criterion: A weighted linear minimum variance approach to combine the estimates from ADS-B and TCAS.

For ADS-B Out, the paper describes implementation on two platforms:

  • Simulation Cockpit Platform: Simulating aircraft signals and transmitting them as ADS-B Out data.
  • Unmanned Aerial Vehicle (UAV) Platform: Equipping a UAV with ADS-B Out capabilities to transmit its own flight data.

Datasets

The paper does not explicitly mention or introduce new benchmark datasets. Instead, it relies on:

  • Simulated Flight Data: Generated using aircraft movement models and flight simulation cockpits.
  • Experimental Data: Collected from the ADS-B Out transmission experiments on the simulation cockpit and UAV platforms.

The "dataset" in this context is primarily the simulated and experimentally generated flight information used to test the developed algorithms and system.

Key Insights and Ideas

  • Fusion is Key: The central idea is that combining ADS-B and TCAS data offers significant advantages over using either system in isolation, particularly in improving the reliability and accuracy of collision avoidance.
  • Adaptive Noise Estimation: The use of Variational Bayesian methods to estimate time-varying noise is a critical advancement, allowing the system to adapt to dynamic conditions and improve tracking performance.
  • Variable Sampling for Robustness: Adapting the sampling rate based on data quality (NACp) makes the system more resilient and efficient, especially when dealing with less precise ADS-B signals.
  • Semi-Physical Simulation: The approach of using a semi-physical simulation environment (combining real ADS-B hardware with TCAS simulation) provides a practical way to validate the fusion system's performance before full-scale deployment.
  • Reduced False Alarms: The fused system demonstrably reduces false and leak alarms, which is crucial for pilot workload and overall flight safety.

Key Results

  • Improved Estimation Accuracy: The fused system's estimation error (measured by RMSE) was consistently smaller than that of individual ADS-B or TCAS subsystems across longitude, latitude, and height. For example, in Figure 10, the "VSampling-IMM-Fusion-RMSE" lines are generally lower than the "IMM-ADSB" and "IMM-TCAS" lines.
  • Reduced False and Leak Alarms: Figure 13 and Table 2 show a significant reduction in TA (Traffic Advisory) and RA (Resolution Advisory) false and leak alarms for the fused system compared to TCAS and ADS-B alone. For instance, RA false alarms dropped from 1104 (TCAS) and 334 (ADS-B) to 172 for the fused system.
  • More Accurate CPA Calculation: The fused system's Close Point of Approach (CPA) time calculation was closer to the true CPA value than either subsystem alone, as shown in Figure 11.
  • Enhanced Safety: The cumulative deviation analysis during RA maneuvering (Figure 14) indicates that the fused system provides a more precise response, contributing to improved flight safety.

Limitations and Future Work

  • Simulation vs. Real-World: While the paper uses a semi-physical simulation, the authors acknowledge that "small deviations" in previous simulations were not sufficient to prove effectiveness, implying that real-world testing is still crucial. The current system is a "minimum system" and may not encompass all complexities of a full-scale operational system.
  • Time Synchronization: The paper mentions that "some methods will be added to consider the measured time which is from the GPS navigation unit to transponder" in future work, suggesting current time synchronization might be a point for further refinement.
  • Model Identification: The authors suggest exploring "pattern recognition or other methods can be used to identify the model which makes it possible to have traffic collision situations," indicating that the current model selection might be improved.
  • Further Fusion Methods: The paper proposes exploring "two-system fusion methods which meet the benefits and security of ADS-B and TCAS fusion," implying that the current VSVB-IMM is one approach among potentially others.
  • Large Data Analysis: The authors suggest future work on "large data analysis of airspace surveillance."

Summary updated

Abstract

Automatic Dependent Surveillance–Broadcast (ADS-B) is the direction of airspace surveillance development. Research analyzing the benefits of Traffic Collision Avoidance System (TCAS) and ADS-B data fusion is almost absent. The paper proposes an ADS-B minimum system from ADS-B In and ADS-B Out. In ADS-B In, a fusion model with a variable sampling Variational Bayesian-Interacting Multiple Model (VSVB-IMM) algorithm is proposed for integrated display and an airspace traffic situation display is developed by using ADS-B information. ADS-B Out includes ADS-B Out transmission based on a simulator platform and an Unmanned Aerial Vehicle (UAV) platform. This paper describes the overall implementation of ADS-B minimum system, including theoretical model design, experimental simulation verification, engineering implementation, results analysis, etc. Simulation and implementation results show that the fused system has better performance than each independent subsystem and it can work well in engineering applications.
BibTeX
@article{wang2017integrated,
  author = {Yanran Wang and Yanran Wang and Gang Xiao and Zhouyun Dai},
  title = {{Integrated Display and Simulation for Automatic Dependent Surveillance-Broadcast and Traffic Collision Avoidance System Data Fusion}},
  journal = {Sensors},
  year = {2017},
  doi = {10.3390/s17112611},
}
Figure