In order to illustrate how the association an d m anagem ent of sensor netw orks w orks, we have developed a d em o nstrator show ing con nected an d associated nodes, the capabilities an d the context data. As a basis, w e use the linked sensor d ata p latform developed at the U ni versity of Surrey w hich is called Sense2Web [7] an d integrated into our connectivity m odule. Sense2Web is a linked-data platform for publishing sensor data an d linking them to existing resources on the sem antic web. As ou r connectivity m odule produces individuals of
the extended W3C SSN ontology it becom es less com plex to connect b o th platform s.
In this prototype w e use O racle SunSpot nodes w ith the deployed m anagem ent m odule to connect to the gateway. The gatew ay h as a SunSpot com patible sink w hich send a beacon signal in a three sec on d interval. The SunSpot can receive the signal an d start to au then ti cate against the gateway. In o u r cu rren t prototype we use sim ple plain authentication an d each n o d e can req uest association by sending its type inform ation, an d the context inform ation described earlier. The gatew ay w ill then create an association response. W hen the n o de as sociates to the gatew ay a tem plate for SunSpot no des is loaded an d ind ividuals are created from the instances respectively. In Listing 1, a snipp et from the ontology is show n in w hich an in dividu al of a node w ith its context inform ation is created.
Listing 1: Instance of Template
< rd f:D escription rdf:about="#SUNSPOT—<xxx).8B54"> <hasSubSystem rdf;resource="#SunTemperatureSensor"/> <hasConnectionInfo rdf:resource = "#responseTime"/> <hasRadioInformation rdf : resource = "#SignalStrength "/> <hasRoutingInformation rdf;resource="#hop_Distance"/> <hasEnergyInformation rdf:resource="#energyLevel"/> <rdf :type rdf : resource="#System"/>
< / rdf: D escrip tion>
< rd f:D escription rdf:about="#SensorContext"> <rdfs:subClassOf rdf:resource = "#System"/> <rdf:type rdf:resource = "#Class "/>
< / rdf:D escription>
SensezWeb links the context inform ation to other d ata sources. The geographical inform ation such as the location of the gatew ay an d the location of the nodes is show n on a m odified google m ap3 as it can be seen in Figure 22.
In an initial evaluation, the softw are w as installed on 8 SunSpot nodes. They w ere activated one after another. The p eriod for association for this nodes w as aro un d 4 m inutes. 145 m essages have been sent includ ing the beacon signal. The overall nu m b er of m essages exchanged w as 177 w ith a size of 4017 bytes an d an average load of 67 bytes p er n o d e d u rin g the transaction process.
3.3.1 Discussion
The m iddlew are com ponent p resented in this section introduces the following novelties: A sensor discovery an d association m echanism th a t enables to autom atically connect from heterogeneous sensor nodes to a gateway. The sensor nodes have to deploy a sm all m anagem ent
3-3 M I D D L E W A R E I M P L E M E N T A T I O N 8 3
Demo
Sensor Map Overlay Information ;
El B E )
a I
I Map I Sareline | Hybrid "j. D iscovered N o d es
S e n s o r d e s c r i p t i o n U R I: sr/spns^Vfiifnrtasf>nsiv;rfy>14 4 0014.4F01.0000.53F6 V 0014.4F01.0000.4BCD S e n s o r t y p e ; f ii u j ./ / d a o & d i a . w o / c l a s s / v a a o .'S il i L o c a ti o n U R I (lo c a l o m o )o g > '): H i B a n d a a o T e i U o e i a t m e S e n s o r L i n k e d - d a t a lo c a t io n U R I: m t D : //d o p g c B a -o fo rr e s o ijr c e rG u ild to fO 4 ^ L i n k e d - d a i s t a g : G e t r e s o u r c e d e s c rip tic k i ( h e r e l G e t n o d e r n io r m a ito n m e r e ) ^ S e n s o r
/
Dfvire and Context Information: Sensor ID: 00i4.4F01J)000.4BCD SensorTvpc; Sun^>ot Bat:er\ intonnaUon: 2Capabilities Temperature S ensor 27,5 Light Sensor
Figure 22: Sense2Web Application
m odule that follows the specifications described above. By following these specifications, an autonom ous discovery and association can be established regardless of the und erlying netw orking protocols. The collected m eta-inform ation from the associated sensor nodes is stored in a sem antic representation. The sem antic representation al lows to create a linked m odel of the nodes. Relationships are form ed by linking nodes th at are in the sam e location, a n d /o r nodes that have the sam e m easurem ent capabilities or observe the sam e phenom ena. In Table 19 a short sum m ary of the features is shown. The features and their selection have been discussed in C hapter 2.
Heterogeneous Node Support A variety of nodes is supported as long as they implement a manage ment module that follows the negotiation and association mechanism. The gateway and the sensor nodes have to communicate over the same network interface type.
Sensor Discovery The discovery of the sensor nodes is supported by providing a auto mated configuration approach, that is able to associate and negotiate the settings between gateway and node without manual interaction. SOA Integration The middleware supports the access to the nodes either via RESTful
services or via a SPARQL endpoint.
Energy-efficient Communication The association and negotiation consider the nodes context informa tion such as battery level and distance to the base station.
Zero configuration The proposed association method does not require manual interac tion, besides the deployment of the management module to the sensor node.
Direct Node Access Nodes and their data can be directly accessed via a Web interface or using the provided services.
Table 19: Features of the middleware component
The m iddlew are, however, at this stage is n ot able to provide con nectivity in mobile scenarios in w hich sensor nodes can move from one area to the other covered by different gateways. This requires a scheme for the proposed association m echanism and a distributed
overlay n etw o rk betw een the gateway. D espite m obility and large- scale n etw ork aspects, the com m unication betw een sensor node and gatew ay regarding traffic optim isation has to be considered. These enhancem ents for the m iddlew are are in tro du ced an d discussed in the next chapter.
E N H A N C E D D A T A C O M M U N I C A T I O N F O R M I D D L E W A R E
In the p revious chapter w e described the developm ent of a novel m id dlew are com ponent for gatew ays for resource constrained devices an d p rop osed a m echanism for node-to-gatew ay association. In this chapter we discuss enhancem ents for the com m unication aspects of the m iddlew are. We introduce su p p o rt for m obile nodes, Gateway-to- Gatew ay com m unication for a seam less com m unication betw een sev eral setups of sensor nodes an d gatew ays by form ing an d introducing a distributed overlay netw ork b ased on the gossiping com m unication m etho d for the m iddlew are com ponent. In ad d itio n w e develop a novel m etho d to reduce the com m unication traffic by sending d ata in tim es of h ig h activity an d aggregate the d ata w h en there is only low activity in the data stream s. We propose a new algorithm called Sen sorSAX th a t extends the Symbolic A ggregate A pproxim ation (SAX) introduced in C hapter 2 to the sensor dom ain.