Research On Key Technology For Ble PEPS

The Passive Entry and Passive Start system based on smartphone is a new trend for the next generation of automotive PEPS.Bluetooth Low Energy PEPS has the potential to replace existing PEPS with its low cost,low power consumption,short delay,and high reliability.Thanks to the wide application of Bluetooth technology,BLE PEPS system leverages user's smartphones or wearable devices as virtual vehicle key which can eliminate specific vehicle key fob and achieve real PEPS.This paper studies positioning-related problems involved in implementing BLE PEPS.Vehicle in-out detection scheme based on Dempster-Shafer evidence theory and distance estimation algorithm are designed.Experiments are conducted to evaluate the performance of the proposed scheme.The major research content includes the following six aspects:1 Experiments are conducted in the vehicular application scenario including BLE RSS temporal and spatial characteristic analysis,influence of common materials on the vehicle,and the influence of the receiver orientation and beacon placement orientation on RSS.2 Moving average filter,Gaussian filter,and Kalman filter are compared on RSS noise suppression.Kalman filter achieves best suppression effect and real-time performance for RSS filtering.3 Based on the experimental analysis of BLE RSS in vehicular and traditional indoor scenario,primary classifiers based on weighted K-nearest neighbor,improved logistic regression and Bayesian probability method is designed.In order to improve the identification accuracy and stability of the system,vehicle in-out detection scheme based on improved DS theory is proposed.4 Considering about the attenuation effect of vehicle metal shell on BLE RSS,a single beacon ranging correction scheme with interactive multi-model extended Kalman filter is proposed.Distance fusion algorithms include extended Kalman filter,Monte Carlo localization and iterative trilateration are compared.5 Different data collection methods in offline phase for vehicle in-out detection including fixed-point data collection,slide rail data collection,ring-area data collection are explored.Suitable beacon placement is also selected by experiments on different candidate positions.6 Static and dynamic experiment are respectively designed to evaluate vehicle in-out detection and distance estimation between user and vehicle.The experimental results show that in simple scenario(BLE RSS are collected only inside the vehicle or outside the vehicle),the average accuracy of vehicle in-out detection reaches 99%,and the static experiment shows the positioning error is within 30 cm.In the complicated scenario which includes vehicle in-out transition,the accuracy of vehicle in-out detection reaches 95%.In some situation,vehicle in-out detection accuracy reaches more than 98%.In the dynamic experiment for distance estimation between user and vehicle in near field range,the probability of the positioning error in the range of 2 meters of the proposed algorithm reaches 90%.