Signature Inspired Home Environments Monitoring System Using IR-UWB Technology

Abstract

Home monitoring and remote care systems aim to ultimately provide independent living care scenarios through non-intrusive, privacy-protecting means. Their main aim is to provide care through appreciating normal habits, remotely recognizing changes and acting upon those changes either through informing the person themselves, care providers, family members, medical practitioners, or emergency services, depending on need. Care giving can be required at any age, encompassing young to the globally growing aging population. A non-wearable and unobtrusive architecture has been developed and tested here to provide a fruitful health and wellbeing-monitoring framework without interfering in a user's regular daily habits and maintaining privacy. This work focuses on tracking locations in an unobtrusive way, recognizing daily activities, which are part of maintaining a healthy/regular lifestyle. This study shows an intelligent and locally based edge care system (ECS) solution to identify the location of an occupant's movement from daily activities using impulse radio-ultra wide band (IR-UWB) radar. A new method is proposed calculating the azimuth angle of a movement from the received pulse and employing radar principles to determine the range of that movement. Moreover, short-term fourier transform (STFT) has been performed to determine the frequency distribution of the occupant's action. Therefore, STFT, azimuth angle, and range calculation together provide the information to understand how occupants engage with their environment. An experiment has been carried out for an occupant at different times of the day during daily household activities and recorded with time and room position. Subsequently, these time-frequency outcomes, along with the range and azimuth information, have been employed to train a support vector machine (SVM) learning algorithm for recognizing indoor locations when the person is moving around the house, where little or no movement indicates the occurrence of abnormalities. The implemented framework is connected with a cloud server architecture, which enables to act against any abnormality remotely. The proposed methodology shows very promising results through statistical validation and achieved over 90% testing accuracy in a real-time scenario.

Keywords: Edge Care System; Indoor Location; Movement Detection; Support Vector Machine; Ultra-Wide Band.

Figure 1

The geometry of azimuth angle.

Figure 2

The secure cloud server and minimal internet of things (IoT) architecture refers to the components of the system embedded within the house comprising a ultra wide band (UWB) data collection front end, storage, and the post processing stages to understand home environment.

Figure 3

P410 device and associated peripheral hardware.

Figure 4

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is present in the kitchen space has been considered as C1 in classification phase.

Figure 5

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is plumping a cushion has been considered as C2 in classification phase.

Figure 6

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is using the microwave in the kitchen has been considered as C3 in classification phase.

Figure 7

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is eating in the dining room has been considered as C4 in classification phase.

Figure 8

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is washing a bowl in the kitchen has been considered as C5 in classification phase.

Figure 9

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is watching television in the living room has been considered as C6 in classification phase.

Figure 10

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is walking from the kitchen through to the dining room and hallway entrance to living room has been considered as C7 in classification phase.

Figure 11

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is brushing their teeth in the bathroom has been considered as C8 in classification phase.

Figure 12

The relationship between propagation delay, activity frequency, and received power from the radar responses obtained while the person is returning from the bathroom to the living room has been considered as C9 in classification phase.

Figure 13

Distance and frequency mapping to agree the floor plan for different categorical events.

Figure 13

Distance and frequency mapping to agree the floor plan for different categorical events.

Figure 14

Scatter plot of categorical UWB localization data.

Figure 15

Confusion matrix.