To ensure that elderly members of society are able to maintain a quality of life that allows them to live independently, it is important that impending frailty, which occurs between the state of begin healthy and needing nursing care, be detectable at an early stage. The purpose of this study was to construct a system that detects pre-frailty, which is the pre-stage of frailty, and notifies the subject's family member or designated watcher. This study focused on the decrease in walking speed of elderly people in pre-frailty, designing a means of measuring walking speed using a non-wearable passive infrared (PIR) sensor that performs equivalent accuracy with conventional manual measurements. The walking speed measured by this system with an average error of 1.63% from the walking speed measured by conventional manual measurement. Further testing confirmed that walking speed could be measured with an average error of 1.51%, even at one meter. The system is feasible, cost-effective, and can be easily installed in homes for continuous measurement of the walking speed of elderly people. In addition, the system's visualization function provides heat maps of the elderly's indoor physical activity, which is effective and useful for allowing caregivers to remotely observe them on the Internet or through a social networking service.