Towards radar-enabled sensor networks

Ultra wideband radar-enabled wireless sensor networks have the potential to address key detection and classification requirements common to many surveillance and tracking applications. However, traditional radar signal processing techniques are mismatched with the limited computational and storage resources available on typical sensor nodes. The mismatch is exacerbated in noisy, cluttered environments or when the signals have corrupted spectra. To explore the compatibility of ultra wideband radar and mote-class sensor nodes, we designed and built a new platform called the radar mote. An early prototype of this platform was used to detect, classify, and track people and vehicles moving through an outdoor sensor network deployment. This paper describes the sensor's theory of operation, discusses the design and implementation of the radar mote, and presents sample signal waveforms of people, vehicles, noise, and clutter. We demonstrate that radar sensors can be successfully integrated with mote-class devices and imbue them with an extraordinarily useful sensing modality.