Simulation Performance Enhancement in Automotive Embedded Control Using the Unscented Transform

Abraham Elias Ortega Paredes, Lauro Roberto Nunes, Max Mauro Dias Santos, L.r.a.x. de Menezes, Kathya Silvia Collazos Linares, João F. Justo, Zonghua Gu

Abstract

Automotive embedded systems comprise several domains, such as in software, electrical, electronics, and control. When designing and testing functions at the top level, one generally ignores the uncertainties arising from the electrical and electronic effects, which could lead to an irregular behavior and deteriorate their performance even using the appropriate methodology for designing the embedded control systems. Then, the studies and comparison on the effect of uncertainty in the automotive domain are important to improve the overall performance of those control systems. Here, we explored the uncertainty in control systems using the Monte Carlo (MC) and unscented transform (UT) methods. These methods have been applied to a mobile seat platform (MSP) and a light emitting diode (LED) used for lighting of heavy-duty vehicles. The UT for embedded control systems has shown better performance when compared to the Monte Carlo method, in order to reduce the number of required variables and computational resources in the simulation of failures and test-case generation. Finally, this investigation brings another application for the UT, in order to exemplify its applicability and advantages when compared with the other methods.

Introduction

In automotive engineering, embedded control systems—such as those managing motor actuation or lighting—are typically designed based on "nominal" values. These fixed parameters assume ideal component behavior. However, real-world electronic components are subject to manufacturing tolerances, thermal fluctuations, and aging, all of which introduce significant uncertainty into system performance.

Traditionally, engineers have relied on Monte Carlo (MC) simulations to account for these stochastic variations. While MC is highly accurate, it is a "brute force" method requiring thousands or even millions of iterations to achieve statistical convergence. This computational burden often makes MC impractical for complex real-time system design or exhaustive automated test-case generation. This paper investigates the Unscented Transform (UT) as a high-performance alternative to MC for modeling uncertainty in automotive embedded systems.

Methodology: Monte Carlo vs. Unscented Transform

The research compares two primary approaches for uncertainty propagation through nonlinear systems:

  1. Monte Carlo (MC): This stochastic approach relies on pure random sampling. To obtain stable estimates of a system's mean and variance, a massive number of individual simulation runs (e.g., 500,000) is often required.
  2. Unscented Transform (UT): UT operates on the principle that it is easier to approximate a probability distribution than an arbitrary nonlinear function. Instead of random sampling, UT uses a small, deterministic set of "sigma points." These points are strategically chosen to capture the mean and covariance of the input variables. When passed through a nonlinear system, these few points provide a high-fidelity approximation of the output's statistical distribution.
The process of combining individual probability density functions (Gaussian and Uniform) into a joint discrete probability density to determine the sum of variables.

Case Studies and Modeling

The effectiveness of the UT method was evaluated using two practical automotive applications based on physics-derived mathematical models:

  • Mobile Seat Platform (MSP): This study modeled the current of a DC motor used in accessibility seating. Uncertainties included input voltage (18V–32V), load torque (9–24 Nm), and motor constants with ±5% to ±8% variability.
  • LED Lighting System: This model focused on the junction temperature of heavy-duty vehicle LEDs powered by a DC-DC Buck converter. Key uncertain parameters included Buck converter capacitance (±10%), inductance (±5%), and thermal resistance values.

Ground truth for these models was established using massive Monte Carlo simulations (500,000 iterations) to validate the accuracy of the UT results.

Results and Discussion

The experimental results demonstrate a significant leap in simulation efficiency without sacrificing statistical accuracy:

  • Accuracy: In the MSP case, both MC (500,000 tests) and UT (729 tests) yielded nearly identical results for permanent current (Mean: ~37.5A, Std Dev: 6.12). For the LED junction temperature, UT achieved the same mean (83.03°C) and standard deviation (4.30) as the full MC suite using only 19,693 tests.
  • Computational Speedup: Depending on the mapping complexity, UT reduced the number of required simulations by orders of magnitude—ranging from 25x to 600x faster than MC while maintaining comparable confidence levels.
  • Sensitivity Analysis: UT facilitates efficient covariance analysis. In the LED study, researchers identified that the Buck converter's output capacitance accounted for 50% of the total variance in junction temperature, providing a clear target for component-level optimization.

Conclusion

The Unscented Transform offers a robust, deterministic, and computationally efficient framework for uncertainty analysis in automotive embedded systems. While it performs best with well-behaved distributions (e.g., Gaussian or Uniform), its ability to provide repeatable results with a fraction of the variables required by Monte Carlo makes it an ideal candidate for integration into Model-Based Design (MBD) workflows. Future work will focus on incorporating UT into automated failure forecasting and real-time test-case generation during early-stage automotive software development.

Summary updated

Abstract

Automotive embedded systems comprise several domains, such as in software, electrical, electronics, and control. When designing and testing functions at the top level, one generally ignores the uncertainties arising from the electrical and electronic effects, which could lead to an irregular behavior and deteriorate their performance even using the appropriate methodology for designing the embedded control systems. Then, the studies and comparison on the effect of uncertainty in the automotive domain are important to improve the overall performance of those control systems. Here, we explored the uncertainty in control systems using the Monte Carlo (MC) and unscented transform (UT) methods. These methods have been applied to a mobile seat platform (MSP) and a light emitting diode (LED) used for lighting of heavy-duty vehicles. The UT for embedded control systems has shown better performance when compared to the Monte Carlo method, in order to reduce the number of required variables and computational resources in the simulation of failures and test-case generation. Finally, this investigation brings another application for the UT, in order to exemplify its applicability and advantages when compared with the other methods.
BibTeX
@inproceedings{paredes2019simulation,
  author = {Abraham Elias Ortega Paredes and Lauro Roberto Nunes and Max Mauro Dias Santos and L.R.A.X. de Menezes and Kathya Silvia Collazos Linares and João F. Justo and Zonghua Gu},
  title = {{Simulation Performance Enhancement in Automotive Embedded Control Using the Unscented Transform}},
  booktitle = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2917931},
}
Figure