II. OVERVIEW OF MotionScale
A. System Overview
In-bed body motion detection can facilitate a variety of
research in Human Computer Interactions (HCI), smart home,
and healthcare, such as home environment control, sleep monitoring,
etc. Our main goal is to detect in-bed body motions by
utilizing low-cost, low-overhead sensing techniques. Toward
this end, we devise a motion detection system based on lowcost
load cell sensors. The system can be easily integrated to
an existing bed by placing the load cell sensors under each
bed leg. The basic idea is to observe the electrical resistance
changes on each load cell to infer possible body motions on the
bed. Intuitively, when a body motion occurs, the body weight
distribution changes, causing each load cell’s resistance to
change accordingly. In this work, we also focus on utilizing the
relative load cell resistance changes to discriminate two types
of body movements: Big Movements and Small Movements.
Big Movements usually happen when there is a motion in the
body’s torso, such as turning to the left or right, and Small
Movements happen when just a small part of the body moves,
such as re-positioning the arm or head. Since our system can
accurately detect in-bed body motions using load cell sensors,
we refer to it as MotionScale.
As illustrated in Figure 1, in MotionScale, each load cell
sends its data using a PIP-Tag (the wireless communication
protocol described in Section III) with a sampling rate of 30
Hz. The base station, which is connected to the USB port of
a laptop, conducts the following processing after receiving the
data:
1) Data Interpolation. We first interpolate the data by
applying the spline interpolation technique