Wireless Sensor Networks


Recently advances in micro-mechanical devices (MEMS) have enabled the development of very small sensing devices called sensor nodes. These sensor nodes are smart devices with sensing, data-processing and transmission (typically radio) capabilities. It is foreseen that the successors of the current prototype nodes are to reach the millimeter scale at low costs (well under a dollar), making it possible to use them in huge ad-hoc networks, called Wireless Sensor Networks (WSNs) reaching magnitudes as high as 10^6 nodes in exceptional cases, and a few orders of magnitude lower in typical cases.

Typical Scenario

Sensor nodes are tiny disposable devices forming dense networks, often called Smart Dust. Such dust clouds are meant to be deployed in an ad-hoc way (e.g. dropped from a boat, a plane, a mortar shell, a space-rocket), making them ideal to be deployed in remote, hostile or difficult to access environments (earth-quake, tsunami, battle-field, forest, planet (space exploration program), sea, river, isolated wildlife-harbor islands), enabling applications as diverse as survivor detection, enemy intrusion detection, target-tracking, forest-fire detection or environmental and biological monitoring. This type of utilisation is not the only one envisaged, for example, patient-doctor health monitoring, blind and impaired assisting, home-applications and industrial supervising applications are also considered. However, different applications may rely on different hardware, i.e. different sensor nodes. In this research project, we have mostly in mind (although not exclusively) application of the area-monitoring type using sand grain like pico-nodes forming a smart dust network. A typical application is the forest-fire detection scenario: a back-pack sized bag of sensors containing 10’000 sensors would be deployed over a forest of approximately 100 [km^2] to detect forest fires in a critical region.

Algorithmic Challenges

Futuristic scenarios like this one and the similar ones envisaged above are realistic under an optimistic but reasonable assumption on the progress of the very active hardware research area of MEMS/NEMS. The foreseen existence of this new ubiquitous type of hardware raises new and challenging algorithmic paradigms, because the standard algorithms and protocols are inadequate, since they do not address the specific requirements imposed by WSNs. The challenges to be overcome range from foundational work (e.g. theoretical modeling and analysing of efficiency, robustness and scalability as well as new and specific distributed-algorithmics and protocols) to more engineering aspects, like middle-ware design, or MAC layer protocols. The classical approaches are essentially inappropriate because of a few characteristics of WSNs which make them very different from traditional networks, like computer networks (e.g. the Internet), ad-hoc or mobile ad-hoc networks (MANETS). Most important differences follow from hardware specificity, which has to fulfil constraints like miniaturization and low-cost of nodes, thus making it possible to employ very dense networks (thus a scale difference with traditional networks), but which also imply the nodes will be limited in available energy, transmission power, computing power, available memory, etc... That is, some resources which are usually available become scarce in sensor networks. Amongst resource constraints, energy scarceness is probably the tightest, and it comes as an essential guideline in the analysis, modeling, study and development of distributed algorithms for Sensor Networks.



In the Battelle Research Center, Building A, we have deployed a wireless sensor network testbed consisting (at the moment) of around 70 sensors with varying capabilities. The sensing capabilities include temperature, luminocity, GPS, cameras, passive infrared and acceletometers. The network spans across the 2nd and 3rd floor of the building and is being used for testing and experiments in wireless sensor network research. The sensors used are made by Coalesenses GmbH and belong to the iSense platform. This wireless sensor network is part of the Wisebed project and is deployed, developed and maintained mainly by Marios Karagiannis.