Espresso: Testing Synthesized Programs of Wireless Sensor Network

Description

Wireless sensor network (WSN) applications are becoming increasingly important. They sense physical events in situ and analyze the sensed data in-network. Examples include hospital administrations and automatic detections of geological events. Many of them run on motes with limited capabilities. In practice, embedded software engineers need to develop these applications by trial and error to fit the resource constraints, for which the process is tedious, time-consuming, and error-prone. Testing is a major means for assuring the quality of software. However, if the above adaptation is done manually, testing all of these trial-and-error versions would be ineffective and inefficient.This project aims to integrate the testing with an automated selection of WSN components to fit the resource constraints. A WSN application is modelled as a set of tasks, in which each task consists of components. The modelling and arrangement of test cases at the component level, and the test execution at the task level, will also be investigated. Since different components of the same task may have non-identical invocation rates, the project aims to investigate the effective and efficient methods to execute test cases across components of the same and different tasks. Feature interactions will also be explored in the project.