Embedded Sensing for Standardised Urban Environmental Monitoring

More documents

Recommendations

Info

A. Usage of Open StandardsThe components of this process chain are separated byseveral interfaces, which are defined using open standards.The first central group of standards is subsumed under theterm Sensor Web Enablement (SWE), an initiative by theOGC that aims to make sensors discoverable, query-able,and controllable over the Internet [16]. Currently, the SWEfamily consists of seven standards, which encompass theentire process chain from making sensors discoverable in aregistry, to measuring physical phenomena, and sending outalerts. [17]• Sensor Model Language (SensorML) – This standardprovides an XML schema for defining thegeometric, dynamic and observational characteristicsof a sensor. Thus, SensorML assists in the discoveryof different types of sensors, and supports theprocessing and analysis of the retrieved data, as wellas the geo-location and tasking of sensors.• Observations & Measurements (O&M) – O&Mprovides a description of sensor observations in theform of general models and XML encodings. Thisframework labels several terms for themeasurements themselves as well as for therelationship between them. Measurement results areexpressed as quantities, categories, temporal orgeometrical values as well as arrays or composites ofthese.• Transducer Model Language (TML) – Generallyspeaking, TML can be understood as O&M’spendant or streaming data by providing a methodand message format describing how to interpret rawtransducer data.• Sensor Observation Service (SOS) – SOS provides astandardised web service interface allowing access tosensor observations and platform descriptions.• Sensor Planning Service (SPS) – SPS offers aninterface for planning an observation query. Ineffect, the service performs a feasibility check duringthe set up of a request for data from several sensors.• Sensor Alert Service (SAS) – SAS can be seen as anevent-processing engine whose purpose is to identifypre-defined events such as the particularities ofsensor measurements, and then generate and sendalerts in a standardised protocol format.• Web Notification Service (WNS) – The WebNotification Service is responsible for deliveringgenerated alerts to end-users by E-mail, over HTTP,or via SMS. Moreover, the standard provides anopen interface for services, through which a clientmay exchange asynchronous messages with one ormore other services.Furthermore, Sensor Web Registries play an importantrole in sensor network infrastructures. However, they are notdecidedly part of SWE yet, as the legacy OGC CatalogueService for Web (CSW) is used. The registry serves tomaintain metadata about sensors and their observations. Inshort, it contains information including sensor location,which phenomena they measure, and whether they are staticor mobile. Currently, the OGC is pursuing a harmonisationapproach to integrate the existing CSW into SWE bybuilding profiles in ebRIM/ebXML (e-business RegistryInformation Model).An important ongoing effort in SWE development is theestablishment of a central SWE Common specification. Itsgoal is to optimise redundancy and maximise reusability bygrouping common elements for several standards under onecentral specification.The functional connections between the describedstandards are illustrated in Fig. 2.Figure 2. Functional Connections between the SWE Standards.Besides these sensor-related standards, other OGCstandards are used for data analysis and provisioning. TheWeb Processing Service, as described in [18], provides aninterface to access a processing service offering a number ofpre-defined analytical operations – these can be algorithms,simple calculations, or more complex models, which operateon geospatial data. Both vector and raster data can beprocessed. The output of the processes can be either a predefineddata structure such as Geographic Markup Language(GML), geoRSS, Keyhole Markup Language (KML),Scalable Vector Graphics (SVG), or a web-accessibleresource like a JPEG or PNG picture.Standardised raw data access is granted by the use ofOGC WFS, WMS and WCS standards. These well-knownstandards provide access to data in various formats such asvectors (points, lines and polygons), raster images, andcoverages (surface-like structures).More about the described sensor related and dataprovision standards can be found on the OGC web site 1 .B. Location-aware Complex Event ProcessingApart from standardised data transmission and provision,a special focus in the Live Geography approach is theextension of Complex Event Processing (CEP) functionalityby spatial parameters. In a geographic context, CEP can forinstance serve for detecting threshold exceedances, for geo-1 http://www.opengeospatial.org
fencing implementations, for investigating spatial clusters, orfor ensuring data quality.Generally speaking, CEP is a technology that extractsknowledge from distributed systems and transforms it intocontextual knowledge. Since the information content of thiscontextual knowledge is higher than in usual information, thebusiness decisions that are derived from it can be moreaccurate. The CEP system itself can be linked into an EventDriven Architecture (EDA) environment or just be built ontop of it. While the exact architecture depends on theapplication-specific needs and requirements, a generalarchitecture including five steps can be applied to every CEPsystem. Fig. 3 shows this architecture in Job DescriptionLanguage (JDL) [19].Figure 3. General CEP Architecture with Levels of Data Processing. [19]The system is divided into several components (or levels)that represent processing steps. Since individual systemsmay have different requirements the implementation depthwill vary from system to system. The following list notes thebasic requirements for each level, according to [19]:• Level 0: Pre-processing – Before the actualprocessing takes part the data is normalised,validated and eventually pre-filtered. Additionally, itmay be important to apply feature extraction to getrid of data that is not needed. Although this part isimportant for the CEP workflow, it is not aparticularity of CEP in general. Thus, it is marked aslevel 0.• Level 1: Event Refinement – This component’s taskis to track and trace an event in the system. After anevent has been tracked down, its characteristics (e.g.data, behaviour and relationships) are translated intoevent-attributes. The tracing part deals with stateestimation that tries to predict upcoming events inthe system.• Level 2: Situation Refinement – The heart of eachCEP system is the step of situation refinement,where the actual analysis of simple and complexevents takes place. This analysis includesmathematical algorithms, which comprise not onlycomputing boundaries for certain values, but alsomatching them against patterns and historical data.The results are high level (contextual) interpretationswhich can be acquired in real-time.• Level 3: Impact Assessment – After analysing andrefining the situation, it is important to generatedecisions that may have consequences. Impactassessment deals with the simulation of outcomes. Ittherefore deals with various scenarios and simulatesthem by accounting for cost factors and resources.Results are weighted decision reports that includepriorities and proposals for corresponding scenarios.• Level 4: Process Refinement – Finally, the last stepcovers the interaction between the CEP system andbusiness processes. It provides a feedback loop tocontrol and refine business processes. While thiscould include integration and automatic controllingof processes, it may also just be the creation ofalerting messages or business reports.Originally, CEP has been developed and traditionallybeen implemented in the financial and economic sectors topredict market developments and exchange rate trends. Inthese areas, CEP patterns emerge from relationships betweenthe factors time, cause (dependency between events) andaggregation (significance of an event’s activity towardsother events). In a location-aware CEP, these aspects getextended by additional parameters indicating locationinformation of the event. Spatial parameters are combined onpar with other relational indicators.As geo-referenced data is therefore subject for furtherprocessing, one important aspect is to control its data quality.Quality criteria comprise lineage, logical consistency,completeness or temporal quality. From a practicalviewpoint, this means that the location may be used to definequality indicators, which can be integrated into CEP patternrules.C. ImplementationThe implementation of the Live Geography approachcomprises tailor-made sensing devices, a real-time dataintegration mechanism, a use case specific interpolationmodel, an automatic server-based analysis component, and acomplex event processing and alerting component. All thesestand-alone modules are chained together using openstandards as described below.For the measurement device, we designed a specialsensing pod for pervasive GIS applications using ubiquitousembedded sensing technologies. The system has beenconceived in such a modular way that the base platform canbe used for a variety of sensor web applications such asenvironmental monitoring, biometric parameter surveillance,critical infrastructure protection or energy networkobservation by simply changing the interfaced sensors.The sensor pod itself consists of a standard embeddeddevice, a Gumstix Verdex XM4 platform including anARM7-based 400MHz processor with 64MB RAM and16MB flash memory. It runs a customised version of the„Open Embedded“ Linux distribution (kernel version 2.6.21)with an overall footprint of
Page 1 and 2: Bernd ReschManfred MittlboeckFabien
Page 3 and 4: This paper begins with a review of
Page 5: • Global Monitoring for Environme
Page 9 and 10: Fig. 6 shows the interface, which i
Page 11: VI. EFFECTS ON THE URBAN CONTEXTFro

Embedded Sensing for Standardised Urban Environmental Monitoring

Create successful ePaper yourself

Delete template?

Save as template?