To handle the over fitting and inadaptability problem of current fall detection algorithms caused by lack of real fall samples of elderly people and the use of small size of fall samples collected by young people, a fall detection algorithm based on random forest was proposed. By adopting sliding window mechanism, the sequentially collected acceleration data within the window were firstly processed to extract feature parameters of time domain and frequency domain, and then the Bootstrap approach was employed to randomly select partial samples with the same number from the whole training sample collection, after that random selection of features was performed to construct a collection of basic SVM (Support Vector Machine) classifiers with best feature partition. On the online fall detection stage, the final classification result was obtained with vote of results by multiple basic SVM classifiers according to the majority criteria. The experimental results demonstrate that the proposed algorithm outperforms the SVM and BP (Back Propagation) neural network algorithms with 95.2% accuracy, 90.6% sensitivity and 93.5% specificity, and reflects that the fall detection algorithm based on random forest can accurately recognize the fall activity, and has strong generalization ability and robustness.

Considering the low power consumption for successive indoor and outdoor scenes pervasive perception in complex and dynamic environment, a lightweight indoor and outdoor scene identification algorithm based on Support Vector Machine (SVM) multi-classifier was proposed, which can accurately distinguish the indoor and outdoor scenes with low power consumption. The algorithm adopted data mining method to obtain different characteristics in indoor and outdoor scenes from the sensors integrated in smart phones (such as visible light sensors, magnetic sensors, acceleration sensors, gyro sensors, and pressure sensors, etc.). It also made advantage of human behavior difference between indoor and outdoor scene. According to different time and weather conditions, the algorithm designed support vector machine multi-classifier to identify complex indoor and outdoor scenes based on the differences of human behavior in indoor and outdoor scene. The simulation results show that the proposed algorithm has good universality, and can determine the indoor and outdoor scenes with more than 95% accuracy, and only consumes less than 5 mW averaging power.