Wireless sensor networks (WSN) [1] are currently an active field of research. A WSN consists of large numbers of cooperating small-scale nodes capable of limited computation, wireless communication, and sensing. In a wide variety of application areas including geophysical monitoring, precision agriculture, habitat monitoring, transportation, military systems and business processes, WSNs are envisioned to fulfill complex monitoring tasks.
In many typical sensor network applications, fine-grained physical locations of individual sensor nodes play an important role. Examples include target detection (where is the target?), target tracking (where and how fast is a target moving?), and target classification (what are size and shape of the target?). Moreover, location-dependent queries can increase both the utility and the lifetime of a sensor network. By directing a query only to nodes in a certain geographical region or by making certain query parameters (e.g., sensor sampling rate) a function of the node location, valuable energy resources can be saved by restricting the sensor network activity to what is actually needed to answer a query.
Techniques for physical location sensing have been studied for a long time, among others, in the context of mobile computing systems [16]. More recently, some of the approaches developed there have been adopted for WSN [3,4,6,10,13,25], mainly focusing on systems based on certain characteristics such as time-of-flight, received signal strength, signal range of ultrasound and radio waves. This adoption is often possible, because in many respects a wireless sensor node is not too much different from a mobile computing device like a PDA with WLAN access. Compared to a PDA, however, sensor nodes have rather limited resources due to their required small size and cost. Nevertheless it is often possible (both energy-wise and size-wise) to equip such sensor nodes with low power radios or small ultrasound transducers as enablers for location sensing systems.
However, research is already on the way to create the next generation of sensor nodes, for example at UC Berkeley [19,31]. Due to their envisioned cubic-millimeter size, they are called ``Smart Dust''. By making nodes inexpensive and easy-to-deploy, Smart Dust opens up new applications areas. The radical size reduction mandates a revolutionary change in the used communication technology when compared to current WSN technology. Traditional radio technology presents a problem because Smart Dust nodes offer very limited space for antennas. Furthermore, radio transceivers are relatively complex circuits, making it difficult to reduce their power consumption to the level required by Smart Dust. In order to meet these requirements, [19] suggests the use of laser-based free-space optical transmission. However, due to power restrictions, near future Dust nodes will most likely make use of passive optical communication only, limiting communication to a bidirectional link between a base station device and each node.
This revolutionary new technology presents a whole new set of challenges to location sensing systems. Traditional systems based on radio waves and ultrasound are ruled out due to their power consumption and size requirements. The expected unprecedented scale of Smart Dust deployments will further challenge a location system.
In this paper, we present the Lighthouse location system for future WSN systems that are similar to the early Smart Dust prototypes developed at UC Berkeley [19]. By extending the base station, this system allows Smart Dust to autonomously estimate their physical location with respect to a base station with high precision over distances of tens of meters without node calibration. Besides a modified base station, the system does not require any additional infrastructure components. This is achieved by a new cylindrical lateration method. In contrast to traditional spherical methods, this approach does not have a wide baseline requirement (see Section 3). On the receiver side, only a simple optical receiver (amplified photo diode), moderate processing capabilities, and little memory are needed. That is, only marginal changes to the Smart Dust prototype developed at UC Berkeley are necessary.
We first describe future Smart Dust systems and compare them to more traditional commercial-off-the-shelf (COTS) sensor nodes. We will then describe the challenges of a location system for Smart Dust, before presenting the Lighthouse location system itself. The latter includes a description of the basic approach, the presentation of a prototype system, a set of initial measurements, and a first analysis of several system aspects. We conclude the paper with mentioning related work, our current work, and future research directions.