IEEE Consumer Electronics Magazine - March 2018 - 75

and ac [13] is used to locate the hidden target by comparing
the signals from the reference and surveillance channels.
However, this application needs specific devices to receive
and calculate the parameters in Wi-Fi signals, so it is not suitable for general-public use.
Another application is a three-dimensional motion-tracking
system [6], which was developed by transmitting a hundredth
of a wireless signal and calculating the x and y dimensions to
localize the human body. This can facilitate different applications, e.g., elderly monitoring, fall detection, and controlhousehold applications. However, this design requires the
generation of a specified Wi-Fi signal by converting the power
to the hundredth scale.
There are several studies on Wi-Fi RSSIs. For example,
[2] describes how the high-Doppler resolution method and
Doppler-only tracking method could achieve indoor tracking using Wi-Fi. The filters used in the Doppler methods
include the extended Kalman filter (which relies on a linear
model) and the SIR Filter (which uses particles for measurement). However, experiments in [2] were only performed on a target walking in a straight line and not in
other scenarios (e.g., moving in a circle, behind obstacles,
or based on hand gestures).
Another study in [7] focuses on the features and principles
of a Wi-Fi fingerprint-based positioning algorithm. This algorithm locates the position of a subject based on the probabilistic fingerprinting method by capturing the signal strengths
of Wi-Fi access points in an indoor area, and the experimental results show that the algorithm has lower power consumption and less computational power. However, the experiments
only focused on power saving, calculation time, and battery
lifetime and not on the detection accuracy. Moreover, detection accuracy is an important factor when choosing a positioning algorithm.
Unlike the presented studies, this article focuses on the
RSSI change under different environment conditions (e.g.,
various types of obstacles and different hand gestures performed). We also use the RSSI change to design and develop
a video player that is controlled by hand gestures. Furthermore, there are three other related works [8]-[10]. Unlike [8]
and [9], which use channel state information (CSI), we rely
on an RSSI. The focus of [9] is on recognizing the English
words by hand drawing in the air, whereas we recognize the
individual hand gestures. Unlike [10], which requires the gestures to be performed closely above the receiver, we do not
limit the movement of the gestures and allow the gestures to
be performed in front of the receiver. In addition, [11] and
[12] are recent works on device-free activity recognition and
human-computer interfaces.

CAPTURING RSSI VALUES
The RSSI of Wi-Fi signals are stored in the radio header of
802.11 wireless local area network packets. To evaluate and
analyze RSSI values, we need to set up an experimental
environment to capture the wireless packets. Moreover,
because the passive system approach is a flexible and

FIGURE 1. The experimental setup.

extendable way to measure RSSI, we used this approach in
our experimental format.

EXPERIMENTAL SETUP
In our experimental setup, we used a broadband router as the
transmitter and a universal serial bus Wi-Fi adapter as the receiver, as shown in Figure 1. A Linux computer was connected to the
receiver for data processing, different kinds of obstacles were
placed between the transmitter and the receiver, and varying
hand gestures were performed between the transmitter and the
receiver. A packet-capturing tool was run on the Linux computer
to capture the wireless packets from the transmitter, and the
RSSI values were then extracted from the packets and analyzed
using other tools. We used tcpdump as the packet-capturing tool,
which is a powerful command-line packet analyzer. However,
the monitor mode had to be configured in the wireless interface.
The interface under monitor mode oversees all the packets
received through the interface, and tcpdump cannot obtain the
RSSI values without a monitor mode interface.

EXPERIMENTAL PROCEDURES
There are five experimental steps.
1) Change the wireless interface card to monitor mode.
2) Collect data packets using tcpdump.
3) Export the RSSI values stored in the 802.11 radio header
to a text file for statistical analysis.
4) Calculate the mean, mode, maximum, and minimum values of the RSSI.
5) Study the RSSI values versus time graph and identify the
features of the RSSI value changes.

EVALUATION OF THE RSSI CHANGES
FOR DIFFERENT HAND GESTURES
EXPERIMENTAL DESIGN
Fourteen hand gestures were performed to evaluate the corresponding change of the signal strength components in the Wi-Fi
RSSI, as shown in Table 1. Each hand gesture was evaluated 30
times, and the red-, blue-, and green-dotted lines represent the
signal strength versus the time graphs of three samples.

RESULTS OF EXPERIMENTS ON HAND GESTURES
Table 2 summarizes the average, maximum, and minimum
RSSI values and the characteristics of the waveforms using
RSSI signal strength versus time graphs.
MARCH 2018

IEEE Consumer Electronics Magazine

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - March 2018