![[Logo] Instituto Superior Técnico](../../img/wwwist.gif)
|
|
URBISNET - Urban pollution monitoring using a public transportation infrastructure for networked sensingProject PTDC/EEA-CRO/104243/2008, sponsored by the portuguese national science foundation (FCT) Status: Started in March 2010 PI: João Gomes AbstractThe main scientific goal of this project is to develop tools for estimating a diffusive field, such as the concentration of gases over a certain area, using a set of mobile sensors. The envisioned application proposed here is pollution monitoring in the city of Lisbon based on measurements taken by the fleet of public urban buses (Carris). This approach could significantly increase the density of sampling across the city with only a modest investment in infrastructure, thus providing a much more detailed picture of air quality than using the current small network of fixed measurement stations. A prototype system will be developed, comprising a few sensing/computing units installed on buses, a central station (CS) where data on the reconstructed pollution field will be made available, and all necessary components to convey readings or computed variables via an ad-hoc wireless network. The latter will link buses and a set of gateway nodes that interface to the internet, through which the CS is reached. The system output will include pollution maps with fine spatial and temporal resolution. This proposal stems from current research activities of the participants on the topic of sensor network (SN) design and monitoring of critical infrastructures. Although interesting problems arise at several levels when designing SN, the main emphasis here is on information processing aspects. Theoretical challenges include reconstructing a diffusive field spatially and temporally from irregular and possibly sparse samples; determining the location of major sources of pollution; developing decentralized algorithms to achieve these goals while harnessing the computing resources available in network nodes; selecting appropriate subsets of sensors for attaining a favorable accuracy/complexity tradeoff in a given inference problem based on acquired data; selecting sampling locations along bus routes, possibly in an adaptive way, to best reconstruct the pollution map with a limited number of data points. Problems related to the design of an ad-hoc network of mobile nodes are also pertinent to develop the envisioned monitoring application. While buses could certainly be equipped with communications hardware to contact one another and/or the CS through standard mobile networks, doing so would impose operating costs that could become significant for large fleets. Instead, it is proposed that buses be fitted with low-power hardware allowing peer-to-peer communication at ranges of a few tens of meters, bypassing the need for mobile networks. Data packets will then be forwarded between buses, if necessary until they can be delivered to a gateway node that interfaces to an existing wired data network. From that point they can be readily delivered to the CS. A small number of such gateways would be placed at strategic points in the city. In this scenario multihop routing of packets is greatly simplified by the fact that bus routes are fully known a priori. Challenges related to the design of such a network include selecting optimal locations for the sinks and the corresponding routing tables for peer-to-peer packet forwarding. Regarding the design of sensing nodes, energy constraints are not as stringent as in other SN were nodes must operate on batteries without servicing for extended periods of several months or even years. Bus-mounted nodes have access to the vehicle's power source, although one can argue that a self-contained package would be more convenient. For that reason the hardware to be developed will emphasize extended autonomy by adopting low-power transceivers for wireless communications, of the type commonly used in other SN (IEEE 802.15). The hardware will include sensors for measuring the concentrations of several gases such as CO, NO2, O3, SO2 and CO2. Know-how for designing such systems exists among project participants. The sensors mentioned above are not ideal, as readings for the concentration of one gas will be affected by the presence of some of the other ones. Moreover, measurements cannot be taken statically as recommended by the sensor manufacturers, and the air flow due to bus motion will bias the sampled values. Then, one of the scientific challenges in reconstructing pollution fields will be to devise a sort of deblurring step to approximate ideal sampling of gases. Solutions to be examined will include local processing at sensing nodes, possibly with limited interaction with neighboring nodes as buses cross, as well as centralized approaches. One of the practical goals of this project will be to make the collected information available to the public. Primarily this will be done by developing graphical representations where pollution data will be superimposed on city maps. Other possibilities, such as feeding a network of informative panels, may be considered. Finally, note that the proposed sensing nodes will contain GPS receivers and therefore the sampling infrastructure could be easily adapted to estimate traffic congestion along bus routes and time of arrival at bus stops. TasksTask 1 - Theory and Methods for Inference of Diffusive FieldsObjectives:
Expected results:High-level numerical algorithms (written in Matlab) that:
Task 2 - Network Planning and Design. Software DevelopmentObjectives:
Expected results:
Task 3 - Mobile Sensing Hardware for Pollution MonitoringObjectives:
Expected results:
Task 4 - Integration, Testing, and DemonstrationObjectives:
Expected results:
|
||