Introduction With the improvement of living standards, owning a private car has become one of the symbols of a family entering a fashionable and modern lifestyle. However, with the increase in the number of cars, thieves have also begun to target cars as high-end consumer goods. Although most cars are currently equipped with factory-installed anti-theft devices, and there are many brands of aftermarket car anti-theft devices on the market, these devices either focus only on passively handling alarm situations or do not pay enough attention to the reliability of the alarm in judging the theft process, resulting in frequent false alarms and missed alarms, causing a lot of unnecessary trouble for car owners. A car anti-theft system based on sensor information fusion technology, based on a comprehensive analysis of various car theft methods, utilizes the differences and complementarities in the performance of multiple sensors, integrates multiple data information from various sensors, and uses a fusion method for data analysis to proactively and accurately determine the status of the car. Furthermore, the system can communicate with the car owner via wireless network to report the car's status in a timely manner, improving the reliability of the alarm system. Structure of the Anti-theft System The anti-theft system based on multi-sensor information fusion adopts a modular structure. Based on selecting the basic modules that ensure system functionality, the number of modules or sensors can be added or removed according to different user needs to adapt to the requirements of personalized products. The system mainly consists of a monitoring module, a central processing module, an alarm module, an execution module, and a communication module, as shown in Figure 1. The monitoring module comprises multiple sensors and signal processing circuits, used to collect vehicle status information and transmit the processed data to the central processing module. The central processing module receives information from various sensors, fuses the data, and classifies the alarms. Based on different alarm levels, the central processing module triggers different alarm signals and operations, and transmits the situation assessment results to the vehicle owner via the communication module. When the situation is deemed severe, the execution module can directly block the vehicle's ignition circuit, preventing the vehicle from starting. The alarm module receives instructions from the central processing module and uses sound and flashing lights to sound an alarm. The execution module prioritizes vehicle owner commands received via the communication module, allowing the owner to easily lock the vehicle, deactivate the alarm, etc. The communication module enables data communication. The system transmits graded alarm information to the vehicle owner via SMS through a GSM/CDMA network to remind the owner to pay attention to vehicle safety; simultaneously, the owner can remotely control the vehicle using SMS messages. To ensure the rational selection of sensor components, the following analysis of car theft methods is presented. Generally, the most common theft methods used by thieves include: ① Vehicle removal: The vehicle is towed away directly using a trailer, or loaded into another vehicle and transported away. This method requires very little time, allowing thieves to quickly leave the scene. ② Vehicle damage: In unattended parking areas, thieves use tools to break the car door locks or windows to force their way into the vehicle and drive it away by connecting it to electricity. ③ Key duplication or lock-picking skills: The key is copied or lock-picking tools are used to open the car door, allowing the thief to drive away directly. ④ Stealing items from inside the vehicle: Thieves dismantle external parts of the vehicle or smash windows to steal items from inside. These actions cause vibrations and tilting of the vehicle, and also emit infrared radiation with a center wavelength of 9–10 μm, which is related to human bio-information. By identifying unusual phenomena caused by car thefts, selecting appropriate monitoring devices for data collection and employing suitable data processing algorithms can achieve accurate early warning. To extract sufficient information and accurately identify the monitored target, sensor selection should adhere to several basic principles: rationally select and optimize sensor combinations to meet the system's high precision and low cost requirements; use different types and functions of sensors to leverage their respective advantages, share information, and reduce the possibility of false alarms or missed alarms; and use multiple sensors of the same type in a reasonable layout to eliminate blind spots and improve system reliability. Based on the above analysis, the following sensors are selected to constitute the monitoring module of the anti-theft system: ① Microwave Doppler sensor. A sensor utilizing the Doppler effect can detect the movement of a person or object. When a person or object approaches, the receiver frequency changes; when the frequency changes to a set value, it can be determined that a person or object has entered the anti-theft system's warning range; ② Vibration sensor. A vibration sensor using an acceleration detection device can monitor vibrations in specific frequency bands of the vehicle body and generate an alarm when the vehicle body is damaged by external forces; ③ Tilt sensor. Tilt sensors monitor whether the vehicle body tilts relative to its initial position. If this angle change occurs at a specific frequency or reaches a set threshold, it can be determined that the entire vehicle has been moved. ④ Pyroelectric infrared sensors. Pyroelectric infrared sensors are only sensitive to infrared radiation with a center wavelength of 9-10μm and can detect infrared information emitted by the human body. They can be used as a detection device for human intrusion into the vehicle. ⑤ Hall effect switches. Hall effect switches can be used to monitor the unauthorized opening of car doors, engine hood, and trunk lid. These sensors monitor changes in wave frequency, vehicle vibration, changes in vehicle position, and human biometric information. By rationally arranging them to eliminate blind spots, sufficient information for vehicle anti-theft can be obtained, minimizing information uncertainty and improving alarm reliability. Multi-sensor information fusion refers to the process of using multiple sensors collaboratively and effectively combining various sensor information to form a high-performance perception system to obtain a consistent description of the environment. Compared to single-sensor systems, multi-sensor information fusion offers significant advantages in reducing information uncertainty, improving the accuracy of information acquisition, reducing system costs, and increasing system response speed. Therefore, information fusion technology can be used to fuse data collected by monitoring modules. Information fusion models can be studied and represented from several aspects, including function, structure, and mathematical models. The functional model describes the main functions, database, and interactions between system components during information fusion, starting from the fusion process. The structural model describes the hardware and software components of the information fusion system, related data flows, and the human-machine interface between the system and the external environment. The mathematical model consists of the information fusion algorithm and its comprehensive logic. The functional model of multi-sensor information fusion can be divided into three levels: detection-level fusion, location-level fusion, and attribute-level fusion. This anti-theft system classifies alarm levels into four categories: whole vehicle handling, vehicle damage, unauthorized opening, and burglary, and uses different information fusion models for each category. Among them, whole vehicle handling, vehicle damage, and illegal opening involve detection-level information fusion, forming a distributed detection system. Each group of sensors monitors targets independently and without conflict, allowing their information to be comprehensively superimposed using a combination method. Vehicle burglary involves attribute-level fusion, requiring joint estimation of the target's identity, which can be achieved using a guided method. In detection-level fusion, the system's structural model adopts a serial structure. That is, based on the judgment results from the microwave Doppler sensor nodes, it is fused with the detection information from the tilt sensor, vibration sensor, and Hall effect switch. The process is as follows: the microwave Doppler sensor's response signal indicates the approach of a person or object. When the approach of a person or object reaches the warning distance, if the tilt sensor determines that the vehicle's tilt angle reaches a set threshold, it indicates whole vehicle handling; if the vibration sensor also senses vibrations in a specific frequency band of the vehicle body, it is judged as vehicle damage; if the Hall effect switch installed on the door reacts, it indicates that the door or front/rear hatch has been illegally opened. Their respective structural forms are shown in Figure 2. The judgment process for vehicle burglary using attribute-level fusion adopts a tree structure. Based on the results of identifying instances of vehicle damage or unauthorized opening, combined with information from infrared sensors, it's possible to determine if someone has entered the vehicle. Conclusion A car anti-theft system utilizing multi-sensor information fusion technology avoids the false alarms associated with using vibration sensors alone and expands the monitoring range compared to using infrared sensors alone. Its monitoring results exhibit high accuracy, reliability, and real-time performance, offering advantages unmatched by traditional car anti-theft systems. Conclusion The monitoring method for car anti-theft systems based on sensor information fusion technology discussed in this paper can also be applied to various scenarios such as community security and home security. This anti-theft method has a very broad application scope and should be one of the development directions for security systems.