05898077

Wireless M2M Communication Networks for SmartGrid Applications

C. Wietfeld, H. Georg, S. Groning, C. Lewandowski, C. Muller and J. SchmutzlerTU Dortmund University

Communication Networks Institute (CNI)Dortmund, Germany

Email: {christian.wietfeld, hanno.georg, sven.groening, christian.lewandowski, christian5.mueller, jens.schmutzler}@tu-dortmund.de

Abstract—Future Smart Grid applications rely on highlyreliable and secure connectivity between various infrastructurecomponents from utilities, grid operators and households in orderto cover requirements on communications for smart meteringand decentralized energy management including electric mobility.Therefore several approaches for communication infrastructuresare currently discussed based on different architectural conceptsincluding wired and wireless access and inhouse communicationtechnologies. In this paper we provide an overview on wireless-enabled networking architectures and discuss two exemplarynetwork planning approaches. On one hand, network planningalgorithms for the neighbourhood area network (NAN) are intro-duced. On the other hand, we present results for an integrated,wireless network coverage-aware planning of public and privatecharge point locations. In addition we address in the paper thedesign and implementation of embedded, wireless Web Services toenable the efficient and reliable data exchange within distributedenergy systems. The paper concludes with the comparison of twoapproaches for the optimal coding of energy-management-relatedSOAP messages.

I. INTRODUCTION

Future Smart Grid applications rely on highly robust andsecure communications between utilities, providers and house-holds in order to cover the demands on smart metering, energymanagement, decentralized energy generation and electricmobility. Therefore several approaches are currently discussedbased on different architectural concepts including wiredand wireless access and inhouse communication technologies[16],[17].

In order to meet the limitations of resource-restrained ICTinfrastructures for large scale roll-outs, two major areas ofinterest are identified in this work:

• Demand-oriented M2M access network planning• Application protocol optimization in resource-restrained

M2M environmentsIn the following section II an overview on the use of

wireless M2M technologies in Smart Grid applications ispresented. Moreover a decentralized system architecture fo-cussing on the ICT integration of households is presentedand discussed with respect to various application scenarios.In section III a method for hierarchical demand-orientednetwork planning is described in which multiple technologiesare combined and real-world measurements are included.Optimizations on the application layer are discussed in section

IV and finally the paper closes with conclusions and an outlookon future work.

II. CONCEPTS AND REQUIREMENTS FOR ROBUST M2MCOMMUNICATIONS IN SMART GRIDS

Various Smart Grid approaches are currently discussed andevaluated in different pilot projects with respect to differentapplication scenarios (see Figure 1). The driving force ofestablishing an ICT infrastructure and enabling enhancedenergy management services is given by smart metering appli-cations, which started to appear for industrial and commercialcustomers in recent years [7]. New solutions based upon theuse of wireless (e.g. UMTS, GPRS [6]), wired (e.g. PowerlineCommunications, DSL) as well as P2P (e.g. Goteborg [8])wide area communication systems enable smart meters totransmit their metering data, receive tariff information andprovide additional information to customers. With DLMS[1], SML [2], M-Bus [3] and the upcoming Smart EnergyProfile 2.0 [18] various M2M interaction schemes are offeredproviding data containers, reliable protocol operation and se-curity mechanisms. Recent efforts in energy efficiency inducedby statutory directives combine smart metering services anddynamic tariffing for multiple metering devices (energy, gas,heating, water, etc.) for private households with less frequentmetering intervals due to data protection regulations. Basedupon this approach and due to different technology lifecyclesnew architectural concepts are introduced by decoupling me-tering devices from ICT technologies with ICT gateways [4]for aggregating metering data and bridging access networks toinhouse networks.

By establishing an ICT infrastructure for metering gateways,enhanced energy management for decentralized power gener-ation (e.g. solar panels, wind power, power-heat coupling) anddemand side management (e.g. household appliances, electricvehicles) are feasible for private households. The integrationof these components together with decoupling of ICT andenergy components comes along with the necessity for reliableinhouse communication networks which increasingly use wire-less technologies for interconnectivity. For this purpose severalsolutions are discussed and evaluated, like wireless M-Bus[4] and KNX-RF [4] for transmission of metering data. Forinhouse energy management and home automation ZigBee,

European Wireless 2011, April 27-29, 2011, Vienna, Austria ISBN 978-3-8007-3343-9 © VDE VERLAG GMBH

Paper 1569422357 275

Neighbour Area Network Home Area Network

Wireless ICT Infrastructures: GSM, LTE, WiMAX

Wired ICT Infrastructures: ISDN, DSL, FTTx, Cable, PLC

Powerline Abbreviations: EU: Energy Utility DSO: Distribution System Operator AGG: Aggregator MRO: Meter Reading Operator BST: Base Station Tranceiver

Prosumers

BST

Energy Trading Markets

Infrastructure Components

DSO TSO

MRO

EU

AGG

Regional Energy Marketplaces

Metering

HAN Gateway

Load Management of

Appliances

Inhouse

Applications

Management

HAN Gateway

Management of

Distributed

Generation

E-DeMa

Marketplace

Meter Reading

Operator

Smart Metering for Electricity, Gas, Water, Heating…

Distribution

Network

Fig. 1. Wireless M2M Communication for Smart Grids

Wireless LAN, narrowband RF systems as well as severalPowerline Communications (PLC) technologies are used.

A further application scenario with additional requirementson the ICT infrastructure is the integration of electric vehicles(EVs) into smart grid infrastructures. For the integration ofcharge spots into the Smart Grid basic functionalities forbilling, accounting and invoicing are required. Here the useof wide area mobile communication modules based on 2G,3G or 4G networks will be deployed. In long term, withan increasing market penetration of EVs the charging powerwill increase especially with DC fast-charging stations beingdeployed. Without the coordination of simultaneously chargingEVs the risk of local substation blackouts increases. Thereforea reliable wide area communication link is an essential corerequirement which deployed system must meet in the mid tolong-term. Based on this idea this work focusses on demand-oriented network planning combined and evaluated by real-world measurements and optimizations of application layerprotocols in resource-restrained environments.

In order to ensure transparent connectivity to all inhousecomponents the comprehensive introduction of HAN Gate-ways is one of the central elements for combining the highdemands on security and providing an extensive connectivityto the prosumers household. On the one hand the gatewaysact as firewalls, on the other hand the gateways provideconnectivity to all HAN entities. Thereby functionalities likesmart metering for multiple metering devices, Demand Side

Management for loads and decentralized energy generation,as well as User Interaction are provided by the gateway.

The HAN Gateway collects and stores metering data fromseveral metering devices, such as electricity, gas, water andheating meters using dedicated wireless technologies likewireless M-Bus, KNX-RF or ZigBee. The collected data isbundled and securely transmitted to the meter reading operatorusing wireless wide area point-to-multipoint technologies, e.g.GSM, UMTS, WiMAX, LTE or via meshing technologiesusing ZigBee. Alternative to these approaches wired solutionsare used for inhouse and wide area communication but are nottaken into the following considerations.

The energy management can be done either through theprosumer itself, motivated by tariff or through the DistributionNetwork Operator for controlled or emergency load reductions.Therefore interfaces to the prosumers appliances, loads andlocal power generation components are provided by integrat-ing these components into the prosumers inhouse network.Depending on an external pricing information several loadscan be controlled, e.g. the charging process of electric vehiclesas well as controlling home heating systems or an intelligentwashing machine makes use of the dynamic tariff informationby starting the washing procedure in low tariff periods andavoids starting it in high level tariff periods. In addition tothis, the connection interface can be used for maintenance,remote configuration issues and firmware updates.

One of the key capabilities is the integration of decentralized

276

(b) Simulation environment

(c) Network analyzer results

Long

itude

Long

itude

Long

itude

(a) Satellite ViewLatitude

Latitude

Latitude0 m

0 m 1500 m

1500 m

0 m

0 m 1500 m

1500 m

0 m

0 m 1500 m

1500 m

Num

ber o

f com

mun

icat

ion

node

s in

rang

e

Household with Smart Meter

Fig. 2. Demand-Oriented analysis and network planning

power generation systems, which will make up a large part ofthe energy generation of future systems. Nowadays an increasein local power generation installation can be observed, e.g.solar panels, wind power plants and combined heat and powergeneration. A central controlled energetic recovery systemis necessary to meet the requirements of the energy gridquality management. Referring to the communication market adedicated infrastructure has to be provided by an operator, that

is also providing management and installation services to theprosumers. In order to maintain a reliable ICT infrastructurebetween the HAN Gateways and the marketplace, a HANGateway Operator provides the reliable ICT components forsoftware updates, administration and configuration issues.

III. DEMAND-ORIENTED NETWORK PLANNING

In the following, we address two different network planningscenarios. In the first scenario, a methodology is introduced,which allows for the comparison of different network planningalgorithms in real-life scenarios to minimize the infrastruc-ture investment. In the second scenario, the identification ofoptimal locations for private and public charging spots isaddressed.

A. Network Planning for Neighbourhood Area Networks(NAN)

In this scenario, the focus is placed on the dimensioning ofthe neighbourhood area network, in which the so-called con-centrators play an important role. The concentrators exchangedata with the Home Area Networks via the ICT gatewayin each home. Concentrators provide wide-area connectivitywith corresponding hardware costs. Therefore the aim of thedifferent algorithms is, to reach all HANs with minimumnumber of concentrators (i.e. 100% coverage). We have in-vestigated different algorithms, such as demand-oriented andcoverage-oriented algorithms. In addition a greedy algorithmhas been refined through a iterative analysis of the positioningof the nodes to achieve a maximum coverage with a minimumnumber of nodes. In the following, exemplary results arepresented:

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 50 100 150 200 250 300

Co

vera

ge

Number of concentrators

Greedy Adding Algorithmus

flächenorientierte Versorgungsdeckung

nachfrageorientierte Versorgungsdeckung

Optimized greedy adding algorithm

Area oriented supply coverage

Demand oriented supply coverage

Fig. 3. Comparison of the automated network planing algorithms

Figure 2(a) shows a satellite image of the reference scenarioand Figure 2(b) is a screenshot of the simulation in thediscrete-event simulation environment OMNeT++. In Figure2(c) the results of the Network Analyzer for the scenario ata transmission distance of 50m and 10m analysis interval areshown. Along with the number of communication nodes intransmission range the color change from dark blue (smallnumber) to dark red (high number).

Based upon this analysis an automated network planningalgorithm solves the optimization problem which is basedupon a set-covering location problem (see Figure 3). Theprocedure in the case of a complete supply coverage iscontinued until all smart meters are provided.

277

B. Wireless network aware planing of electric vehicle infras-tructures

Next to the simulation of network availability real worldmeasurements at specific positions for the inhouse ICT in-frastructure and charge spots need to be accomplished. Thischapter introduces a measurement environment for analyzingthe availability of 2G and 3G cellular networks. It can beapplied in both the inhouse energy management domain andthe electric mobility domain. As already mentioned in theprevious chapters electric mobility also needs to be integratedin intelligent smart home ICT infrastructures (see Figure 1).The roll-out of electric vehicles (EVs) will rise in the mediumterm and charging infrastructures are needed in the homeenvironment because most of the EVs will be charged atnight when the prices are low. For the charging processof an electric vehicle a long range communication interface(2G or 3G) is installed in current charge spots (CS) forcommunication with the e-mobility-Hub (EM-Hub) a centraldata platform for authentication, invoicing and billing. Due topotential radio reception problems in garages and basementgarages we propose to use the long range communicationinterface provided by the HAN gateway infrastructure toestablish this reliable link in home environments. This linkcan be realized e.g. with DSL but also with wireless wide areacommunication technologies. As the metering infrastructureis in most cases located in the basement of houses the radioreception problems can appear in this case as well and needto be avoided by measurements before the installation. Whenthe reliability of this link is secured an integration of the CSin the HAN can be accomplished using different short rangecommunication technologies, e.g. Powerline Communications(PLC) or wireless technologies of the IEEE 802.11 family orIEEE 802.15.4.

Fig. 4. UMTS measurement for potential home charging spots

Figure 4 shows an exemplary measurement of 3G UMTSnetwork availability in a garage. An attenuation of minimum15dBm can be seen in the whole spectrum. Therefore a UMTSconnection between the CS and the EM-Hub cannot be reliablyrealized using UMTS in this environment. This measurementwas accomplished with a handheld spectrum analyzer FSH8

and an isotropic antenna TS-EMF of Rhode & Schwarz. Thisequipment is also content of the measurement setup developedfor cellular network availability measurements in public andsemi-public areas which is shown in Figure 5. In these areasthe HAN gateway infrastructure is not available and 2G, 3Gor in the future 4G cellular networks need to be utilized. Toensure the connectivity of the CS and therewith the reliablecommunication to the EM-Hub an infrastructure plan for 2Gand 3G networks in Berlin is created in the research projecte-ikt [5]. For this infrastructure plan measurements of GSM(900 & 1800 MHz) and UMTS (2,1 GHz) are accomplishedat potential positions for public and semi-public CS analyzedby project partner TU Berlin.

MySQLDB

USB

Measurement Equipment Control & Evaluation Equipment

CNI Measurement Software ESRI ArcMap

Fig. 5. Measurement environment for cellular network availability

The measurement setup consists of Measurement Equip-ment as well as Control & Evaluation Equipment. The Mea-surement Equipment contains the before mentioned spectrumanalyzer, antenna and additionally a GPS module. With thismodule the exact position of the measurement point can bedetermined. The measurement reports and the GPS positionwill be passed via USB to the CNI Measurement Softwareon the Control & Evaluation notebook. The measurementreports are stored in a MySQL database and consists ofmaximum values of GSM900, GSM1800 and UMTS spectrumin dBm as well as screenshots. For the reason that thesespectrum measurements do not contain provider informationa GPRS/UMTS USB modem is installed to get additionalinformation like the signal strength, cellID and connectionmode for a special provider. With these information thespectrum can be mapped to the frequencies of the networkproviders. For consolidation and visualization of the collecteddata ESRI ArcMap is used in combination with detailed mapsprovided by the administration department of Berlin.

Figure 6 shows the visualization of the communicationinfrastructure plan in Berlin with several measurement pointsaround the potential CS position in Berlin. It can be seen, thatthe radio reception already differs in a small local change.Therefore real world measurements are important to enablethe reliable communication between different entities of thesmart grid.

278

GSM 900 GSM 1800 UMTS100 m

80m

Fig. 6. Visualization of measurements for potential public electric vehiclecharging spots

IV. WIRELESS M2M APPLICATION LAYERCONSIDERATIONS FOR SMART GRID INFRASTRUCTURES

The integration of Smart Grid components in existing in-ternet infrastructures plays a major role for sustainable andcost efficient deployment of the Smart Grid. Within thisbackground being cost efficient and compatible is a balancingact between using efficient communication on ressource con-strained links and for being compatible, using state of the artweb technologies, which has been designed for workstationsund broadband internet connections. At this, finding an optimalbalance is a major challenge, which can be achieved byoptimizing the radio protocol or increasing the efficiency atthe application layer. For that and with respect to the identifiedapplication scenarios in section III-B, this section focuses onincreasing the application layer efficiency by using state ofthe art Web Service-based service provisioning and efficientdata transport on the application layer for wireless M2Mcommunications. [15]

Interoperability and loose coupling of autonomously actingservices are just two prominent characteristics of Web Serviceswhich make them a perfect fit for the integration of SmartGrid components. Both characteristics are achieved throughthe strict adoption of open and standardized internet protocols.

A. Scalable and adaptive Service Provisioning for ElectricMobility

As previously stated in section III-B for the service provi-sioning two domains can be identified. First the inhouse energymanagement and home automation domain which is limitedto a minor number of devices and bounded to one IP subnet.The second domain combines a major number of devices likeall public charge spots of a city connected by long rangecommunication technologies (e.g. LTE, UMTS or GPRS).Furthermore a combination of these two domains is alsoconceivable. For example inhouse components or charging of

electric vehicles in private IP subnets can be securely activatedor deactivated by the external Smart Grid infrastructure forenergy demand clearing through the aforementioned HANgateway.

Web Services are concrete implementations of Service Ori-ented Architecture (SOA) [13] principles. The SOA paradigmis often also referred to as the SOA triangular involving ser-vice providers, service brokers and service consumers. SinceWeb Services are initially evolved as interoperable businessenvironments, Web Services were predominantly deployedas static, always available services. This changes with theemergence of services adopting the Devices Profile for WebServices (DPWS) [10] specification including the Web Ser-vice Dynamic Discovery (WS-Discovery) [11]. Target serviceslike inhouse energy management and home automation aredynamic and might underlie a higher rate of change. Thischaracteristic of target services also implies complex enhance-ments to the Service Broker being the responsible entity forservice discovery. Typical standards and implementations likeUDDI do not handle highly dynamic service availability.

The initial ad-hoc discovery mechanism of DPWS is con-strained to local area networks since multicast messages areused for discovery of new devices and services. These mes-sages are not supposed to pass subnet boundaries in standardnetwork environments and therefore are especially qualifiedfor the inhouse domain. The WS-Discovery specification isan approach driven in conjunction with DPWS to handleservice discovery in dynamic environments. The specificationproposes self management in local environments and involvescentral service brokers the discovery proxies (DP) for widearea environments. The self management in local environmentsis important for a high acceptance of future Smart Griddevices. It brings USB like plug and play functionality to theIP networking domain. The DP as a central service broker istailored towards the second, more scalable domain and pro-vides means to combine both domains. According to the natureof the used wireless cellular wide area networks the discoveryis expanded to more than one subnet and may be combinedwith a Network Address Translation (NAT) [12]. The use ofa discovery proxy and unicast instead of multicast messagesprovides a suitable solution for this challenge. Furthermore,due to the multicast suppression the use of a discovery proxycauses significantly reduced network latency and also reducesnetwork load.

Hence DPWS including WS-Discovery extended with theDP functionality is a potential technology to cover serviceprovisioning in the identified domains for future Smart Gridsincluding electric mobility applications.

B. Efficient Data Transport for Web Services using μSOA

One of the major principles of Web Services is interoper-ability. Hence Web Services heavily rely on W3C’s XML asa core language in order to provide information between het-erogeneous communication entities. As Web Services initiallyevolved in the internet for business environments, the primarytarget platforms were servers and workstations.

279

As described before in section IV-A, a current trend isbringing the benefits of Web Services to the embedded domain.This is primarily reasoned due to the fact that embeddedsystems and especially mobile devices like smart phonesand handhelds become less resource constraint with everyyear and therefore suit the requirements for interaction withweb services or even hosting their own. However, there isstill a need for supporting very constraint, low-cost devicesand also bandwidth limited communication channels (e.g.ZigBee), as they exist in wireless M2M networks and will bepredominantly used in large scale Smart Grid deployments.

Lots of research has been spent in reducing the amountof data to be transmitted over the communication channel bycompressing the XML payload of the messages (e.g. GZIP,WBXML). But this approach is not sufficient in case of somewireless M2M applications since the applicability to resourceconstrained embedded systems for both processing messagesfor serialization / deserialization and the resulting message sizebeing reduced in an easily adoptable and revertible way, mustbe proven. The approach in this work involves an intermediateμSOA Proxy Service. This μSOA Proxy Service performstranscoding between typical Web Service based SOAP mes-sages and purged binary μSOA messages.

The μSOA Proxy Service is an infrastructure service devel-oped for the EU project MORE (Network-centric Middlewarefor Group communication and Resource Sharing across Het-erogeneous Embedded Systems), which has been an Europeanresearch project focusing on the development of an OSGibased middleware using DPWS based communication betweenembedded devices for realizing multiple Web Service basedservices, the so called MORE services. The μSOA ProxyService is a MORE service providing:

• Transparent Proxy Support: SOAP enabled proxy servicerepresenting μSOA enabled devices as standard DPWScompliant endpoint.

• Reduced Messaging Overhead: By utilizing the μSOAProtocol the overhead on the communication channelbetween two nodes is drastically reduced.

These two options extend the Discovery Proxy previouslydescribed in IV-A with typical proxy functionality as it can befound in generic web proxys. Hence, a mirrored DPWS serviceendpoint can be provisioned and also accessed with or withoutadditional message reduction methods through this extension.When using the message reduction, the proxy translates eachmessage between the sender and receiver from SOAP to μSOAor vice versa.

The message transcoding is based on the WSDL descriptionfile of the target service, which forwards the definitions tothe Proxy Service at the beginning of a service interaction.Based on the structure of the WSDL, the μSOA proxy servicebuilds a binary message schema for all possible messageexchanges specified within the WSDL for the target service.The implementation for the target service and the adoption ofthe binary message structure are also automatically built basedupon the given WSDL. The μSOA proxy service is built uponthe MORE Core and adopts the OSGi lifecycle management

provided by the MORE middleware. Hence a service descrip-tion forwarded to the μSOA Proxy Service initiates an on-the-fly creation of a mirrored DPWS compliant service endpointon the proxy, allowing other clients or services to interact withthe target service through standard SOAP interaction with theDPWS service endpoint created on the proxy server.

Adoption of the μSOA principles allows a message sizereduction of at least 70% (in some cases even 98%). Thisleads to a major decrease of communication latency since boththe transmission delay due to the limited bandwidth of thecommunication channel as well as the processing needed forserialization and deserialization are reduced.

The impact of blown up XML-based SOAP communicationon embedded systems and wireless communication links is nota new topic. To counter the overhead problem of XML, W3Cspecified the WAP binary XML (WBXML) which is a binaryrepresentation of XML and therefore much more efficient withrespect to the information per byte ratio. Both standard XMLand WBXML are used for the performance evaluation of WebServices.

A comparison of the average message size transmitted usingWBXML and μSOA compared to an equivalent XML messageis shown in figure 7. The result confirms the completelydifferent approach between the two compression methods:Because WBXML generates an additional, static overheadcompared to the existing XML structure, the compressionrates increases while raising the content of a message. Thisis founded by the increasing number of elements transmittedin the example message. At the beginning the additionaloverhead has a big quota on the message size transmitted usingWBXML and the average message size commutes in about80%. While increasing the transmitted elements, the quotaof the additional header reduces compared to the compressedXML structure and the average message decreases to a sizeof 44.4%.

0%

20%

40%

60%

80%

100%

0,69 0,74 0,83 1,02 1,40 2,16 3,68 6,72 12,79 24,95

Com

pare

dto

SOA

P si

ze

SOAP Size [kByte] μSOA WBXML

Min 45,4%

Max 28,4%

Fig. 7. Average message size of WBXML and μSOA compared to SOAP

The compression ratio of μSOA shows a completely dif-ferent behavior. Due to the fact that μSOA uses the servicespecifications defined in the WSDL, there is no need foradditional header information and so only the actual values aretransmitted. This leads to an oppositional behavior comparedto WBXML and achieves the best compression rate while

280

transporting only a few elements (about 1.2% of the equivalentSOAP message, as seen in 7). Raising the content using μSOAresults in an anti-proportional increasing message size. This isdue to the fact that more elements are transmitted within amessage, the less is taken part by the removed XML syntaxinformation compared to standard SOAP.

In comparison, both approaches reach a point of saturation,at which μSOA shows a rising logarithmic behavior whereasWBXML shows a falling logarithmic behavior.

V. CONCLUSION

This paper presents a general technical overview consid-ering demand-oriented communication networks in terms offuture Smart Grid applications. At this, especially methodsfor planning and optimizing wireless communication networksand achieved results are presented.

Current challenges and possible application scenarios inthe field of Smart Grid have been shown. An overview ofapplicable state-of-the-art wireless technologies for realizingSmart Grid applications is provided and results in the pre-sented architecture for realizing a Smart Grid. These effortsare currently worked out in ongoing research projects likeE-DeMa and e-IKT. In order to provide means for costoptimization of the deployment phase a simulation and real-world measurement based network planning methodology isintroduced. From the first results it can be concluded, thatmeasurements at dedicated positions for HAN gateways andcharge spots need to be accomplished in order to ensure areliable communication.

Furthermore the paper presents first results of the μSOAmessaging framework providing optimizations on web servicecommunications in the application layer. Future work willinvestigate μSOA’s performance compared to other currentstate of the art approaches like Efficient XML Interchange(EXI) [14].

ACKNOWLEDGMENT

The work in this paper was partly funded by theGerman Federal Ministry of Economics and Technology(BMWi) through the projects E-DeMa (reference number01ME08019A) and e-IKT (reference number 01ME09012).The authors would like to thank the project partnersRWE, Miele, SAP Research, Siemens, ProSyst, SWK,Ewald&Gunter, ef.Ruhr and TU-Berlin for fruitful discussionswithin the projects.

REFERENCES

[1] IEC 62056-42: Physical layer services and procedures for connection-oriented asynchronous data exchange

[2] SML - Smart-Message-Language. Version 1.03, November 2008[3] EN 13757-4:Communication systems for meters and remote reading of

meters - Part 4: Wireless meter readout (Radio meter reading for operationin the 868 MHz to 870 MHz SRD band.

[4] Multi Utility Communication, FNN, Version 1.0, August 2009[5] J. Schmutzler and C. Wietfeld, ”Analysis of Message Sequences and

Encoding Efficiency for Electric Vehicle to Grid Interconnections”, IEEEVNC 2010, New Jersey, USA, December 2010

[6] Evaluating The Leading-Edge Italian Telegestore Project, presentation byFabio Borghese, ENEL, Business Development Executive, Infrastructureand Networks Division

[7] Automated Metering and Information System (AMIS), Siemens[8] T. Arnewind, Goteborg - The first ZigBee City?, ZigBee Alliance Open

House, Paris, Mai 2007[9] Wolff, A.; Michaelis, S.; Schmutzler, J.; Wietfeld, C.; , ”Network-centric

Middleware for Service Oriented Architectures across HeterogeneousEmbedded Systems,” EDOC Conference Workshop, 2007. EDOC ’07.Eleventh International IEEE , vol., no., pp.105-108, 15-16 Oct. 2007

[10] Alain Regnier; Dan Driscoll; Antoine Mensch, ”Devices Profile for WebServices Version 1.1,” OASIS Standard, July 2009

[11] Toby Nixon; Alain Regnie; Vipul Modi; Devon Kemp; , ”Web ServicesDynamic Discovery (WS-Discovery) Version 1.1, ” OASIS Standard, July2009

[12] Srisuresh, P.; Egevang, K., ”Traditional IP Network Address Translator(Traditional NAT),” Request for Comments 3022, January 2001

[13] Ken Laske; Jeff A. Estefan, Ken Laske; Francis G. McCab; DannyThornton, ”Reference Architecture Foundation for Service Oriented Ar-chitecture Version 1.0”, October 2009

[14] J. Schneider and T. Kamiya, ”Efficient XML Interchange (EXI) For-mat 1.0,” W3C, Tech. Rep., December 2009. [Online]. Available:http://www.w3.org/TR/2009/CR-exi-20091208/

[15] C. Wietfeld, J. Schmutzler and C. Muller, ”A Smart CommunicationInfrastructure for Future Energy System Applications”, InternationalWorkshop on Future Internet of Things and Services - Embedded WebServices for Pervasive Devices in conjunction with the Future InternetSymposium 2009, Berlin, Germany, FIS, Sept 2009.

[16] C. Wietfeld, C. Muller, J. Schmutzler, S. Fries, A. Heidenreich andH.-J. Hof, ”ICT Reference Architecture Design based on Requirementsfor Future Energy Marketplaces”, First IEEE International Conferenceon Smart Grid Communications (SmartGridComm 2010), Gaithersburg,Maryland, USA, IEEE, pages: 315 -320, Oct 2010.

[17] B. Lichtensteiger, B. Bjelajac, C. Muller, and C. Wietfeld, ”RF MeshSystems for Smart Metering: System Architecture and Performance”,First IEEE International Conference on Smart Grid Communications(SmartGridComm 2010), Gaithersburg, Maryland, USA, IEEE, pages:379 -384, Oct 2010.

[18] ZigBee Smart Energy Profile 2.0 [Online], Available: http://zigbee.org

281