Privacy protected life-context-aware alert by simplified sound spectrogram from microphone sensor

Abstract

This paper introduces a design of life-context-aware alert system from multiple small microphone sensors at various places in home. In order to support the comfortable daily lives of elderly people who live alone, it is important to know their daily activities in home without privacy exposure. In the case of their emergency appeared from overwatching data, the system must alert the situation to the hospitals, ambulances, or their families. To reduce data for fast calculation on PIC and to protect their privacy, the system adopts simplified sound spectrogram from each installed microphone modules. The system first analyses these multiple signals to roughly understand what situation occurs, and decides what type of daily-life are found. When the user's life shows emergent situations, the system alerts to the appropriate contact person or institution. This paper especially describes how to simplify the raw data from the microphone sensor with using frequency/time domain for reducing the amount of data and for privacy protection.