Management support and CPU scheduling performance enhancement in wireless sensor networks

This thesis is focused on issues relating to the management of the operating systems in wireless sensor networks. It deals with the questions like: Why the management components are required in operating systems for sensor nodes? What services could be managed in the operating system for such platforms? What are the management constraints of sensor nodes? TinyOS is one of the mostly used operating systems in this area. TinyOS features multithreading architecture, a very flexible networking stack, and virtual machine implementations. In this thesis, we demonstrate a new scheduling algorithm for TinyOS which especially takes routing issues into account. The algorithm increases the sensitivity and responsiveness in wireless sensor networks. We have modified the priority scheduling algorithm which solves the well-known problem of overloading in simple FIFO scheduling. This scheme improves the fairness of the distributed sensing in the network particularly when the network is subject to mobility, to higher sensor network density and to changes in the topology. In this thesis, we also present the design and the implementation of a new management tool for wireless sensor network. This tool provides management system framework for the operating system components on sensor nodes. It also provides performance management by providing easy means to observe parameters whenever it is required to test dynamic changes on the nodes. The tool also analyzes the obtained results in humanunderstandable and user-friendly way which is highly demanded for WSNs. Moreover, configuration management support makes our tool suitable for providing methods to interact with the system running on the sensor nodes.