Background - Smartphone-Based Human Activity Recognition - page 18

affordable and therefore suitable for commercial use and integration in portable
devices. Noise is however a limiting factor of these sensors which we deal with
through a few signal processing solutions (Refer to Sect. 4.3.2 ) .
Smartphones are one of the portable devices users spend more interaction time
with while performing their usual Activities of Daily Living (ADL). This aspect
combined with their characteristics and the possibility of collecting and distributing
usage data make them a exploitable tool for HAR, for example, when used as wear-
able sensors. However, disadvantages related to these sensors (Sect. 2.3.2 ) such as
obtrusiveness and user-awareness are reduced to some extent taking into account that
smartphones are becoming part of our environment and their users are familiarized
with them.
2.4.3 Opportunistic Sensing
Our environment is now a highly instrumented infrastructure composed of all kinds
of devices and sensors which can provide, to some extent, information about it and
its interaction with users. When these elements are exploited as a sensing resource
and its use is different from its original function, we refer to the term Opportunistic
Sensing . For example: the detection of the activation of light switches at home can
be a good indicator of the location of its occupants and can contribute to infer their
performed activities. Even more, this can be improved if the gathered information
is combined with the one coming from household appliances such as microwaves,
blenders, computers, phones, etc. (e.g. if the lights and the TV of the living room
are on, there is a high chance that somebody is watching TV) (Ogawa et al. 2002 ).
Clearly, advantage is being taken from these existing elements even though neither
the switches nor the appliances were firstly designed for this purpose.
Similarly, smartphones are equipped with a varied range of sensors which can be
exploited for new opportunistic sensing applications. The Nokia N95 mobile phone
for example, was launched in 2009 and it was one of the pioneers in introducing
motion sensors (Campbell and Choudhury 2012 ). Its embedded accelerometer was
originally used only for photo orientation and video stabilization but a subsequent
software upgrade provided developers control over the sensor through an Application
Programming Interface (API) and allowed them to take advantage of it for novel
applications. Even microphones that were primarily designed to make and receive
telephone calls, are now exploited for several mobile applications such as in voice
recognition as interface between the user and the device, and also to measure social
isolation based on the duration of ambient conversations (Lane et al. 2012 ).
Also, the combination of accelerometers and gyroscopes have brought many
advantages regarding spatial positioning and its use on smartphones. The sensing
of 6-axis motion have been already exploited for a richer gaming experience. For
example, in first person shooter games, players can explore three dimensional loca-
tions in a more realistic manner by tilting or twisting their devices instead of using
the touchscreen or other phone controls.