![](bg4.png)
Sender Server Receiver
Log on
Log on
ACK
ACK
Telling the peer logs on
Telling the peer’s UDP
socket address.
ACK
Data Request
ACK
Sending Data
Log off
ACK
ACK
Start
Start
End
End
Sending Data
Go to Start
Telling the peer logs off
Start
Repeat
Figure 4: Overview of the communication protocol
(Sashima and Kurumatani, 2016).
4.3 Data Representation
To share the sensing data between the devices in the
service, the data has two data representations for the
service models: a) “raw data representation” which
includes raw sensing data and b) “abstract data repre-
sentation” which includes the data analyzed at a wrist-
watch device connecting to wireless bio-sensor.
4.3.1 Raw Data Representation
This representation includes raw data obtained by the
bio-sensor for the application that requires to know
detailed data of the sensed user. We assume that typ-
ical usage of this representation is for medical ap-
plications, such as remote monitoring of electrocar-
diograph. Thus the communication is one way from
a sensed user to a care giver, such as medical staff.
This raw representation data is synchronously sent at
a constant sensing interval. For example, if the sam-
pling period of the sensor is 5 msec, the data is sent at
each 5 msec interval.
4.3.2 Abstract Data Representation
two way communication assumed in “P2P telecare
service for family members” requires more compu-
tational power for handling traffics. We have devel-
oped the abstract representation to reduce the traffics
in the two way communication. The representation
includes an abstract data summarized by a data se-
quence of raw sensing data: a “posture” data of the
abstract representation is derived by analyzing a se-
quence of 3-axes acceleration data; and a “heart beat”
data is derived by analyzing a sequence of electrocar-
diographic data. In addition, the abstract represen-
tation data is asynchronously sent when the system
detects a change of the status. For example, the data
is sent as soon as possible when the posture of the
user is changed. Hence, using this representation, the
amount of traffic data becomes 10–100 times smaller
than using the raw data representation in our proto-
type system.
4.4 User Interface
The wristwatch type device is tactilely attached with
the user’s wrist and stay close with the user at all
times. To utilize the features of the device, we have
designed a user interface which directly and succes-
sively provide telecare service to the user. The user in-
terface visually shows the users’ current body move-
ments and enables the user to tactilely feel other user’s
heart beats.
4.4.1 Graphical User Interface for Showing
User’s Status
Because the wristwatch-type device has a small dis-
play, we have designed a simple graphical user inter-
face showing the status of the remote user. Figure 5
shows display images of the user interface. In the fig-
ure, each image shows a physical status of the remote
user. Currently, the system classifies the following 6
statuses: standing, moving, laying-up, laying-down,
walking, unclassified. The system automatically clas-
sifies the status based on the 3-axis acceleration data
obtained by a wireless bio-sensor. The sampling rate
of the data is 100 Hz. For the classification, we have
used a rule-based algorithm using predefined thresh-
old values.
To share the classification with the raw data repre-
sentations, the devices share the raw acceleration data
and do the classification based on the data indepen-
dently. Because the classification process is done by
each device, the analysis results can be slightly differ-
ent between the devices.
To share the classification with the abstract data
representations, the devices connected to the bio-
sensor analyses the raw acceleration data and send the
abstract posture data to other devices. Because the
classification process is done by a device, the same
analysis results are shared by the devices.
4.4.2 Tactile User Interface for Feeling Heart
Beats
Based on electrocardiographic data obtained by the
wireless bio-sensor, the system automatically detects
the pase of heart beats of sensed user. To detect
the pase of heart beats, the system monitors peak lo-
cations of the R-wave by calculating differentiation
values of the electrocardiographic data, and detect
sharply changing points as the peaks of the R-wave.
HEALTHINF 2017 - 10th International Conference on Health Informatics
432