Augmented state extended Kalman filter

Augmented state extended Kalman filter (ASEKF) is an advanced estimation technique that incorporates unknown disturbances or biases into the state vector, allowing the filter to adaptively estimate and compensate for system uncertainties. Unlike standard EKF, which assumes zero-mean white noise, ASEKF tracks the actual-value-of-attack or bias in real-time. This capability is critical for compliance with ISO/SAE 21434 Clause 10 (Cybersecurity Design), which requires automotive systems to be resilient against malicious data manipulation. The technique enables the detection of sensor-level attacks by monitoring the residual-adjusted innovation sequence, providing a mathematical foundation for robust anomaly detection in connected and automated vehicles (CAV).

In automotive cybersecurity risk management, ASEKF is applied through a three-step framework: 1) Modeling the augmented state-space including control inputs,-states, and attack-related parameters; 2) Real-time residual generation used by anomaly detectors (e.g., OCSVM); 3) Dynamic-gain adjustment to maintain control stability during attacks. For example, a Taiwanese Tier 1 supplier implementing ADAS could deploy ASEKF to detect GPS spoofing or CAN bus-based sensor-spoofing attacks. This application can reduce the-attack-impact-on-safety-risk by up to 35% and improve compliance with ISO/SAE 21434 by providing verifiable evidence of control-loop resilience during cybersecurity incident-response-testing.

Taiwanese enterprises typically face three challenges: 1) Computational constraints on automotive ECUs, which can be mitigated by optimizing the augmented matrix-size or using simplified-order models; 2) Lack of cross-disciplinary talent (control theory + cybersecurity), requiring investment in upskilling or partnerships with specialized consultants; 3) Regulatory uncertainty as standards like ISO/SAE 21434 and UN R155 evolve. The recommended solution is to adopt a phased approach: start with high-risk functions (e.g., autonomous steering), validate with simulation-based testing (SIL/HIL), and then scale to full-vehicle-level implementation within 12-18 months，ensuring compliance with both European and Asian regulations.

Winners Consulting Services Co., Ltd. specializes in Augmented state extended Kalman filter for Taiwan enterprises, delivering compliant management systems within 90 days, with over 100 successful projects. Free consultation: https://winners.com.tw/contact