From conception and for decades, a cell phone was used solely for wireless communication purposes. Recently however, over the past nine years after the Android(TM) and iOS(TM) operating systems were released to the market, its definition has changed. With increasing capabilities, importance of data generation, data collection and processing functionalities, a cell phone has evolved into a mobile smart device e.g. smartphone. Smart devices are emerging into new roles such as a portable computing devices, sensor hubs and Internet access terminals.

As embedded technologies and systems advance, not only smartphones, but all commercial smart devices, such as wearables, smartwatches or head mounted displays, extend with these capabilities. Amongst all, the function of contextual sensing is unique on a mobile smart device more than on any other commercial computing platform. Mobile smart devices are carried in close proximity of users, traveling with them throughout a day sensing what the user experiences. Ambient and on-body contexts are shared with the user and hence the sensing data can accurately reflect an individual's real environment better than any other computing or sensing device. It is likely that most domesticated living beings i.e. humans, pets, livestock, etc. would be associated with a mobile smart device in the future.

As the Internet of Things (IoT) wireless capabilities become more cost effective and are connected to more objects, pervasive deployment can be realized and hence becomes an important information source in contextual sensing. More important, as IoT wireless items can be equipped with various sensing techniques, such as geofencing data or information acquired from any kind of sensor attached to them e.g. temperature, force, strain, pressure, etc..., the sensing result is comprehensive and highly configurable which is impractical for traditional sensors. The following three challenges are the major causes of limited IoT contextual sensing in smart devices.

First, if implementing such sensing capability on a traditional smart device platform, its high current drain becomes intolerable. Second, prevailing smart device platforms are not able to accommodate all IoT contextual sensors and their requirements. Third, there is no solution for the smart device to schedule sensing tasks from different IoT contextual sensors and pre-process sensing raw data at the system's low layer. To conquer these three problems, and the goal of the thesis is to research, design and implement a novel platform in between a smart device's system's hardware layer and operation system layer to accommodate IoT contextual sensors and conduct always-on sensing tasks.