Analysis And Improvement On Task Scheduling Strategy Of Wireless Sensor Networks Operating System

With the rapid development of the MEMS technology, sensor technology, wireless communication technology, the micro-sensor node which have the capabilities of perception, computation and communications begin to emerge. after being deployed into the environment, they can form a self-organizing wireless sensor network. Wireless Sensor Network can be used to detect environmental information actively ,so it has a very broad application prospects. Wireless Sensor Networks are a new technology of acquisition and processing, and one of the hot spots in computer network research.In Wireless Sensor Networks, operating system of wireless sensor network, as the necessary software support, have the responsibility to carry out more complex scheduling and management. It is running in each network node, and it's the foundation of the top applications and protocol operation. There have already been a number of WSNOS(wireless sensor network operating system), such as TinyOS, Mantis OS, SOS, Contiki operating system etc. As an integral part of the core operating system , scheduling system provide scheduled object definition, implementation of reasonable scheduling algorithm, as well as scheduling concurrency control mechanism between objects, and to a large extent ,its scheduling policy determines the operating system performance.TinyOS ,which is developed by the University of California, Berkeley through Smartdust project, is a lightweight open-source embedded operating system . Because of its small size, the structure of highly modular, component-based architecture, low power consumption, etc., it is widely used in WSN and become a lot of system reference design. TinyOS use FCFS (first come first service) task scheduling strategy. in spite of its easy implementation, the FCFS scheduling strategy only consider the arrival time of each task, only in favor of CPU busy type task, and is not conducive to I/O peak-type task, moreover, the strategy can not guarantee that the wireless sensor node give priority to implement the important task in the heavier load situation.To be directed against the deficiency that the TinyOS FCFS and some kinds of static priority based scheduling strategies have, a new TLScheduler based on multi-level task queues was proposed and implemented on TinyOS . It classified tasks into three different types: real-time tasks, non real-time tasks which were further divided into two kinds : important tasks with initial priority and normal tasks. Meanwhile, three different scheduling stratgies were adopted by these three task queues to achieve a special performance. In the end , the results of the simulation indicate that ,TLScheduler can bring system better real-time and throughput performance.