Design And Implementation Of A Multi-core Wireless Sensor Node Operating System

Wireless sensor network (WSN) consists of distributed wireless sensor nodes which monitor the environmental conditions (temperature, sound, pressure, etc.) and send the collected data cooperatively through the network to a main location. In the past years, WSN technology has been used widespread in diverse application domains, including the precision agriculture, the industrial control, the military surveillance, the medical care, the smart home, etc.One challenge of the WSN research is to run the diverse and complicated WSN applications on the resource-constrained WSN nodes. To address this challenge, an operating system (OS) dedicated to the WSN node is needed. With the WSN OS, not only the constrained platform resources can be utilized efficiently, but also the complicated WSN applications can be served soundly.Currently, many WSN operating systems (TinyOS, Contiki, SOS, mantisOS, openWSN) have been designed and implemented. These OSes have different features and are suitable to be used for different application contexts. In TinyOS, the component concept is used and the OS memory cost is low. In SOS, system services and applications are built into the modules which can then be updated dynamically. In mantisOS, the multithreaded scheduling scheme is used and the WSN applications can be executed concurrently by threads. In open WSN, a complete network stack is implemented. With this stack, the WSN technology and the IoT (Internet of Things) technology can be well combined. Although the development of these OSes have prompted the proliferation of the WSN, many challenges still exist in the WSN OS research, such as the achievement of the real-time scheduling on the resource-constrained WSN nodes, the prolonging of the lifetime of the WSN nodes, the improvement of the WSN reprogramming performance as well as the improvement of the WSN system reliability.To address these challenges, a new real-time, reliable and user-friendly WSN OS LiveOS is designed and implemented. LiveOS uses the new research concepts and techniques to address some key research challenges in the current WSN OS. First, both the software technology and the multi-core hardware technology are combined in LiveOS. By doing this, several advantages can be achieved:node energy cost can be decreased efficiently; WSN nodes can become context aware; node reliability can be improved; Moreover, a new WSN OS debugging approach can be achieved. Second, some new design concepts are used in LiveOS, e.g., the hybrid scheduling (bot the event-driven scheduling and multithreaded scheduling) and the shared-stack multithreaded scheduling (rather than the independent-stack multithreaded scheduling) are implemented in LiveOS. By so doing, memory cost of the real-time OS is decreased, and this makes the real-time OS suitable to run on the resource-constrained WSN nodes. In addition, a new WSN middleware EMIDE is implemented and embedded inside LiveOS. With the EMIDE, not only the application programming complexity can be decreased, but also the application reprogramming performance can be improved.By implementing the above mechanisms, LiveOS has the following features:1) It has low memory and energy cost, thus it is suitable to be used on the cheap and small form factor WSN nodes.2) It supports real-time scheduling, thus the real-time WSN applications (smart care, industrial engine control, etc.) can be supported.3) It has low energy cost, high reliability, and can support efficient over-the-air reprogramming, thus it is suitable to be used on the nodes which work outdoor in the harsh environments. Due to these features, it is expected that LiveOS can contribute significantly to the development of WSN OS.