Abstract
Many wireless sensor platforms have been built to date, but few metrics or benchmark suite exist for their quantitative evaluation. Sensing accuracy is an important issue for developers of WSN applications. However, many existing solutions for WSN applications are optimized for ad hoc network research while paying little or no attention to the sensing aspect. We believe that in order for wireless sensor networks to become a commercially successful field, and for WSN developers to choose their right solutions, the sensor platforms must be evaluated by quantifiable metrics. In this thesis a system-level evaluation method and some metrics are proposed. With this approach, users can obtain the system performance rather than resorting to data sheets of components. A signal emulator is designed to address the issue of data controllability and reproducibility. The experiment results compared with gold-reference instruments show that the signal emulator help users obtain reasonable and meaningful evaluation results with the proposed evaluation methodology. This is a first step towards a systematic evaluation for existing sensor platforms. Based on this signal emulator, more functions can be integrated into the evaluation instrument and the evaluation for sensor platforms can be more complete and meaningful.