Instrumentation & Measurement Magazine 24-6 - 52

tag placement and selection to determine an optimal tag array.
Fine-grained postures such as standing, sitting, lying and falling
were tested, each in several different ways, and the system
achieved an accuracy of over 90%.
Human activity can also be predicted and not just recognized.
In [23], a novel framework called RF-ARP is presented.
It consists of three stages: recognition after the activity, recognition
in progress and activity prediction in advance. The
approach in the first stage is recognition of the interaction between
human and devices that could be a channel to recognize
the activity. RFID passive tags are attached to ubiquitous objects
used by the user to detect the interaction between the user
and the objects. For this purpose, a long-distance antenna and
an UHF RFID reader are chosen to cover a large area and scan
all of the installed tags. By this method, activities are detected,
making it possible to record what users have done. In health
assessment, this scheme makes it feasible for the patient to
receive the relevant services according to the system's predictions
in addition to activity detection.
Wi-Fi
Similar to RFID, Wi-Fi based devices also use RSSI. But, as described
earlier, RSSI fails to provide sufficient robustness in
complex indoor environments because it is the superimposition
of multipath signals with fast changing phases [5]. This is
a fundamental problem. Wi-Fi systems can further access the
channel state information (CSI). The fundamental advantage
of CSI over RSSI is the former's ability to resolve multipath
via frequency diversity. In fact, the wireless propagation channel
property can be described by CSI measurements [5]. CSI
of multipath signals is illustrated in Fig. 5. Both RSSI and
CSI have high dependency on the physical layout of the environment;
therefore, adjustments are needed to match the
environment conditions.
In [14], a Wi-Fi based fall detection system utilizing CSI
is proposed. This system is implemented with only one antenna
at the transmitter side and two antennas at the receiver
side. According to this work, raw amplitude and phase information
are not directly usable, while variance of phase
differences over a pair of receiver antennas is a salient feature
for fall and fall-like activities. In addition, in the event of a fall,
a sharp power decline pattern is found in the time-frequency
domain which is used for accurate fall detection. Another approach
with Wi-Fi is proposed in [15] which monitors human
respiration even if the subject is far away from the Wi-Fi transceiver
pairs. The strength of this approach is that it uses the CSI
ratio from two antennas instead of the raw CSI from a single
antenna which leads to noise reduction and, consequently, an
increase in the range of sensing. Also, the use of two antennas
makes it possible to use the phase value along the amplitude to
better cover blind spots. According to the paper, the proposed
system can monitor respiration up to 8 meters away.
Microwave Doppler Radar
Doppler motion sensor is another solution for contactless activity
detection. The well-known doppler effect is used in this
approach to measure object velocity. At first, the sensor emits
an RF wave with a specific frequency or a fixed range of frequencies
to the target. The emitted wave will then be reflected
after hitting the object which could lead to alterations in frequency
based on the velocity and direction of dynamic objects.
As a result, the direction and velocity of the target object could
be estimated by measuring the reflected signal's frequency. A
radar sensor does not have sensitivity to environmental conditions
and does not need direct line of sight.
Radars can operate in two modes: Continuous Wave (CW)
and Frequency Modulated Continuous Wave (FMCW). As depicted
in Fig. 6a, in CW mode the radar radiates waves with a
specific frequency. If both the target object and the transceiver
are fixed, the frequency of the reflected wave is the same as the
radiated wave. However, frequency of the reflected wave increases
or decreases due to the object getting closer to or farther
from the sensor, respectively. Since there is no timing information
available, only the speed of the target object can be
extracted from the continuous wave signal.
In contrast to the CW mode, in FMCW the continuous wave
is modulated in frequency, as shown in Fig. 6b. As a result,
based on the time shift of the received signal with respect to the
transmitted signal, the distance to the object can be measured
in addition to speed information [24].
Fig. 6. Radar in: (a) CW; (b) FMCW mode [24].
52
IEEE Instrumentation & Measurement Magazine
September 2021
Instrumentation & Measurement Magazine 24-6

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 24-6
