Research has developed numerous systems and technologies for automatically locating people, equipment, and other tangibles. [16] gives an excellent overview and taxonomy of such location systems. These systems all involve gathering data by sensing real-world physical quantities. The data is in turn used to compute a location estimate. Common systems use diffuse infrared light [26,27,29], visible light [8,9,33], laser light [22,23], ultrasound [12,13,15,24,25,30], and radio waves [2,3,4,18].
Some systems have been specifically designed for use in multi-hop wireless ad hoc and sensor networks [3,4,6,10,13,25] and do not require any external hardware infrastructure besides the nodes of the network itself. Other systems rely on an external infrastructure typically consisting of many devices, which have to be carefully placed in the environment of the objects being located [2,15,24,29,30].
However, the special characteristics of future Smart Dust systems as described in Section 2 and the resulting requirements for a location system as described in Section 3 rule out the usage of all of these location systems. The small size and limited resources rule out systems based on radio waves and ultrasound, since transducers for these physical media are too large and transceivers consume too much energy for Smart Dust nodes [19]. The employed optical single-hop communication rules out all systems which require neighbor-to-neighbor or multi-hop communication. The use of a single (or few) base station(s) rules out all system which require substantial external infrastructure. The required localized location computation (see Section 3.2) rules out all systems where nodes cannot compute their location on their own. Additionally, many systems (e.g., ones based on vision and broadband ultrasound) typically have a high processing overhead and large memory footprint due to the necessary signal processing on the raw input data (e.g., time series of images or audio samples).
The systems that come closest to fulfilling the requirements of Smart Dust are ones based on vision or laser ranging techniques. Laser ranging systems are based on measuring the distance between the laser ranger device and some passive object by a variety of different methods [35]. However, with all these methods, only the active laser ranger can estimate the distance, not the object being located, which precludes localized location computation. The same is true for vision based methods, where a high resolution video camera is used to estimate node location [8]. There are also systems which combine laser ranging and vision-based methods [23], which obviously suffer from the same problem.