2012 3rd IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), Berlin

Device-to-Device Communication for Smart Grid

Devices
Simon Fey, Pascal Benoit, Gregor Rohbogner, Andreas H. Christ, Christof Wittwer

Abstract—This paper analyzes the applicability of existing Many other applications are conceivable. Applications that
communication technology on the Smart Grid. In particular it belong to different domains (e.g. transmission, distribution,
evaluates how networks, e.g. Peer-to-Peer (P2P) and decentralized markets, customers) evince different communication and infor-
Virtual Private Network (VPN) can help set up an agent-based
system. It is expected that applications on Smart Grid devices will mation structures [2]. To merge these different communication
become more powerful and be able to operate without a central infrastructures, domain independent middleware that ensures a
control instance. We analyze which requirements agents and fast, secure and stable communication and provides interfaces
Smart Grid devices place on communication systems and validate for numerous plug and play feasible devices and applications
promising approaches. The main focus is to create a logical is required. This paper focuses on finding a solution to
overlay network that provides direct communication between
network nodes. We provide a comparison of different approaches reach “network interoperability” considering the GRIDWISE
of P2P networks and mesh-VPNs. Finally the advantages of mesh- Architecture Council (GWAC) Interoperability Stack [3]. In
VPN for agent-based systems are worked out. particular, this paper examines several well-known techniques
Index Terms—Communication networks, Network topology, for building up a logical network towards the applicability in
Smart grids the Smart Grid solutions. We will discuss several problems
which we expect to occur when applying state of the art
I. I NTRODUCTION communication techniques (e.g. VPN or P2P) to Smart Grid
applications.
T HE share of fluctuating renewable energy generation has
significantly increased over the last years. The electrical
system requires generation and consumption to be equal at
This paper is organized as follows. Section II introduces
an agent-based control system for a Smart Grid and points
any point in time. In conventional energy system power out the need for decentralized communication and direct
plants control their generation depending on the demand of information exchange between devices. In Section III we
the consumers and ensure the compliance of this physical define our understanding of the term Device-to-Device (D2D)
conditions. In contrast, renewable generators can only produce and how logical overlay networks help in providing a virtual
power when natural resources are available. This makes the network on top of a heterogeneous network. The required
balancing more challenging. features that a communication network has to provide to
Enhanced consumers and generators are necessary, to be able fulfill the requirements of an agent-based control system are
to react to local and global grid situations. These sophisticated pointed out in Section IV. Section V then compares VPN and
devices in combination with information technology is what P2P as two different approaches to create an overlay network
experts usually call the Smart Grid. and provides a comparison on several projects that implement
Many different approaches exist how this Smart Grid may be those concepts. In Section VI we conclude our results on
realized. Such new approaches can be clustered with respect decentralized communication networks and their applicability
to their realized “control structure“ or “control concept” [1]. to the Smart Grid.
Common to all Smart Grid applications is the need of a reliable
and secure underlying communication infrastructure.
II. AGENT- BASED S MART G RID
For instance, a home energy management (HEM) system
might communicate via the Internet with the control center This section will outline an application much discussed in
of a Virtual Power Plant (VPP) in order to influence the former and current times: agent-based energy management
energy consumption of the building. In this case the two structured in a fully distributed manner and following a grid-
devices which are communicating are the HEM and a server oriented approach [1], [4], [5].
in the control center. In this scenario the HEM will most In an agent-based Electrical Power System (EPS) different
likely use the Asymmetric Digital Subscriber Line (ADSL) or computer systems, distributed in the electric grid and each re-
Cable Internet connection of the household. In another Smart sponsible for the control of an underlying physical system, are
Grid application a charging station of an electric vehicle (EV) interconnected via a heterogeneous communication network.
receives energy tariffs via GSM to charge the battery with On every device one or more software agents are running.
respect to the electricity grid. Such systems are called Multiagent Systems (MASs). Figure 1
depicts a MAS following a fully distributed control structure.
S. Fey, A. H. Christ are with Offenburg University of Applied Science, Within this example the MAS is responsible for controlling
Germany, e-mail: simon.fey@hs-offenburg.de
P. Benoit, G. Rohbogner and C. Wittwer are with the Department Smart the voltage level in a low-voltage grid while trying to allow a
Grid, Fraunhofer Institute for Solar Energy Systems, Freiburg, Germany maximum of renewable energy generation.

978-1-4673-2597-4/12/$31.00 ©2012 IEEE

 2

Since there is no central control unit for the voltage-grid, authentication is mandatory.
energy agents (A, B and C in Figure 1) have to cooperate and
negotiate when, how and at what intensity each is allowed The initialization of communications between two agents
to use the energy grid. Such an application requires an requires an overlay network which provides the following
appropriate D2D communication, which we will illustrate in functionality and offers an independent interface for any
the following paragraph. application assigned to the Smart Grid:
In our example, agents are located in the same low-voltage 1) Find other agents in the communication network after
grid, willing to cooperate for the purpose of voltage control. first plug-in (Bootstrapping)
The agents differ in the communication transmission mediums 2) Provide information about where detected agents are
they have at their disposal and in the underling physical located in the electric grid and which services they offer
system for which they are responsible (e.g. a combination (Search methods)
of a Combined Heat and Power (CHP) and thermal storage). 3) Ensure fast and reliable communication between agents
Here agent A and agent B have a conventional ADSL con- (Self-healing, Connectivity)
nection whereas agent C uses a Global System for Mobile 4) Ensure secure data exchange and authentication (Secu-
Communications (GSM) connection to the Internet. rity)

GSM
III. D EVICE - TO -D EVICE C OMMUNICATION
Communication between two devices is a commonly used
technique occurring billions of times in every second [6].
GSM Millions of switches and routers, providing the infrastructure
of the Internet, basically exchange data between two devices.
Radio transmitters were already able to communicate between
two devices one century ago. Cell phones communicate with
WWW central stations. These stations are routing calls between the
ADSL two subscribers. Every single transmission takes place between
ADSL
two devices.
This raises the issue about the term D2D in this paper. The
purpose of this section is to define D2D communication and
to distinguish it from the cases explained above.

Agent Agent GSM A. Definition of Device-to-Device communication

A B C We define D2D communication as a direct communication
Agent
between two applications, located in different nodes of a
dedicated logical network, distributed over a heterogeneous
Fig. 1. Direct Communication Between Energyagents via a Heterogeneous
underlying network.
Communication Network

The first steps which are necessary before agent B is able

to negotiate with agent A and C after connecting to the ADSL
router are:
1) Figure out how to connect to other agents in the Internet:
At first, the system or agent needs to figure out information
about how to connect to other agents in the Internet. This
service can either be provided by a central server, or in a
peer-to-peer manner, as we propose in chapter V.
2) Find the appropriate communication partner: Since the
agent B is only interested in communicating with agents which Fig. 2. Heterogeneous Underlying Network
offer a certain service or/and are located in the same low-
voltage grid, the communication system must provide further 1) Underlying Network: Figure 2 points out the
information about the offered service and the position of the heterogeneous underlying network. The term heterogeneous
agents within the electric grid. refers to separated networks and that the technologies and
After finding the appropriate communication partner, the network protocols might be different. The underlying network
agents need to synchronize their system times, e.g. to ensure includes a complete protocol stack responsible for physical
that they negotiate within the same time slots. Furthermore transmission over the medium and routing of Protocol
it is necessary to ensure that third parties cannot listen to Data Units (PDU) between two end-devices. It offers the
the communication or manipulate the negotiation process. possibility for applications to establish connections to any
For that reason an encrypted data transfer with a reliable other application on any other node.
 3

as Network Address Translation (NAT). By default, routers do

not allow incomming connections if they haven’t been initiated
by a device from the internal, local network.

A. Self-Healing
In the area of small-scale decentralized energy systems it
is expected to have high fluctuations of devices. Since most
of today’s available monitoring and control systems rely on
Fig. 3. Direct Communication Between Logic Nodes Over a Hybrid GSM/GPRS or other telecommunications links such as ADSL
Underlying Network
or Cable the link quality may change over time. Aside from
that, systems may not be supplied by power all-day and might
The small red circles and connections added in Figure 3 switch off when their service is not needed. Therefore the
illustrate D2D communication based on this heterogeneous network has to provide self-healing mechanisms that automat-
underlying network. It forms a new logical network of a ically reconfigure the network to achieve scalability and handle
subset of nodes (Figure 4). disappearing and appearing nodes.

B. Bootstrapping
When a new device or agent is powered on and plugged
to the Internet it will need to gather information about other
network nodes. This process is known from P2P networks and
in the following sections will be defined as bootstrapping. This
process includes steps that a new or existing node is taking to
find network nodes and connect to the network.

Fig. 4. Logical D2D Network C. Security

In general, communication in Smart Grid requires user man-
2) Dedicated Logical Network: According to the require-
agement to add and remove devices from the communication
ments identified in Chapter II, Smart Grid applications, first of
network and prevent unauthorized access. Hence, security in
all, need to reach other devices suitable for their application
this perspective is defined as authentication/authorization and
over the underlying heterogeneous communication network.
traffic encryption.
Thereby the number of nodes in the underlying heterogeneous
network might exceed the number of Smart Grid devices
by several orders of magnitude. Hence, it is reasonable to D. Connectivity
concentrate on this subset of nodes in further considerations.
This subset forms a new dedicated logical network between Connectivity in the following defines the ability to set up
those nodes hosting Smart Grid applications. It is an overlay a connection between two applications. Because the Internet
network spread over an underlying heterogeneous commu- is separated in local networks and public domain, it is a key
nication network. The logical network might be established requirement for a communication system to connect devices
directly between applications like, for instance, P2P networks from all kind of networks together. First of all it is checked
do. Or it can be based on virtual network links like VPNs if the examined communication solution can make devices to
work. In this definition the dedicated logical network includes connect at all by, for instance, using a relay in the overlay.
both. It contains all subscribers of one Smart Grid application. Secondly it is checked if a direct communication link can be
established using NAT traversal.
IV. R EQUIRED COMMUNICATION FEATURES
The requirements that decentralized operating devices place
E. Storage and search of information
on communication systems have already been mentioned in
Section II. In this chapter they are pointed out in more detail In a decentralized network the organization of information
and defined as options to check the evaluated systems for. For is challenging as it gets lost with disappearing nodes.
this purpose we are assuming the as-is situation in today’s Searching information is a key feature that a decentralized
Internet. The Internet is a heterogeneous network with lots network will have to provide for agent systems to find data or
of nodes with static and dynamic network addresses. Most of services from other agents. A lot of effort has been invested in,
the devices that use the Internet actually are hidden behind a for instance, filesharing systems to ensure the availability of
router or firewall which translates local network addresses to information and their retrieval in highly flexible environments.
public addresses and back. This proceeding is generally known
 4

V. C OMMUNICATION N ETWORKS nodes and loosely coupled communication for a lot of appli-
Two different communication network types are considered cations. They also provide efficient mechanisms to store and
for decentralized, direct D2D communication. First is solving search data decentralized in, for instance, Distributed Hash
the problem of connectivity on the application layer with Tables (DHTs).
P2P protocols and second is the use of VPN which maps
a heterogeneous to a homogeneous network. Both methods B. Mesh-Virtual Private Network
provide their own advantages and disadvantages.
Mesh-VPN is characterized by more than one network node
Application layer based solutions can be customized for a
that provides packet encapsulation and routing functionality
certain use-case and are independent of underlying systems.
for any kind of application protocol. Mesh-VPN is abandoning
Skype, for instance, solves the challenge of connectivity
the typical client-server structure known from centralized
on the application layer and successfully establishes
VPN by adopting mechanisms from P2P networks. Mesh-
an independent P2P network structure in heterogeneous
VPNs create an overlay network with virtual topology
networks. Unfortunately, it is a closed system. This network
upon existing homogeneous or heterogeneous network. The
cannot be used for other purposes than those it has been
organization of an EPS with overlay networks has already
originally designed for.
been proposed in [10]. In [9] it is examined, how P2P
overlays and mechanisms improve VPNs to easily create
Network layer solutions provide a higher flexibility than
and maintain groups. A high grade of self-organization and
application layer based methods, since network and appli-
automatic configuration is confessed for mesh-VPNs in [8].
cation are separated from each other. Although most VPN
In Table I mesh-VPN networks are analyzed in more detail.
solutions are static and require a lot of configuration, VPN
is widely spread and commonly used for communication
1) tinc-VPN: tinc [14] is an open source project that allows
between nodes in heterogeneous networks. A major advantage
to create a full-mesh VPN across the Internet. According
of VPN technologies is that, due to routing over a VPN server,
to its feature to forward packets between nodes from the
hosts transparently may connect to every other host. Also
entire network, tinc is ideal to set up site-to-site VPNs which
authentication and encryption is shifted to the lower network
means linking several Local Area Networks (LANs) to an
layer. The encapsulated application protocol is not responsible
even bigger network. Tinc requires every connection to be
for security and has not to be aware of threads from outside
manually configured and included to the network with private
the tunnel. Centralized VPN systems are state of the art for
and public keys on every node.
remote monitoring and control of field devices, e.g., a VPP
[7]. Since we are focusing on decentralized applications a
2) P2PVPN: P2PVPN [15] is an open source Java
centralized communication system is not intended. In [8] and
implementation of a full-mesh VPN. It uses BitTorrent
[9] several decentralized, so called mesh-VPNs are examined
trackers for bootstrapping and allows to dynamically integrate
concerning their design and automatic configuration. Hence, in
or remove nodes. P2PVPN tries to directly connect with
the following two subsections P2P networks and mesh-VPNs
every network node and supports relaying packets if nodes
are investigated in detail for agent-based systems.
are unreachable behind a NAT.

A. P2P Networks 3) cloudVPN: cloudVPN is a very similar networking

One possible technology that could be used for commu- approach to tinc. It also features NAT traversal but differs from
nication in Smart Grid are P2P networks. P2P networks for cloudVPN in packet forwarding and site-to-site functionality.
Smart Grid have been considered in [10]–[13]. P2P networks
set up an overlay network on the application layer which 4) Hamachi: Hamachi [16] has become famous in the
means that addressing, routing and network management are field of online computer games as it is a very user friendly
handled by the application. Two famous representatives for tool to create a virtual LAN across the Internet. Hamachi is
P2P networks are Gnutella and BitTorrent. Both networks based on a central service hosted by LogMeIn to bootstrap
have been designed for file-sharing between peers without the network. NAT traversal also allows to set up a direct
having a client-server infrastructure. Although we know that communication link in a virtual overlay network.
those protocols will not serve for setting up an agent-based
Smart Grid out of the box, they are mentioned because 5) N2N: Network to Network (N2N) [17] spans up a
they provide several interesting mechanisms for decentralized virtual layer 2 network across separated networks. Nodes
communication. Concerning the requirements listed in Section initially connect to a supernode and may set up direct links
IV, it is examined which mechanisms P2P networks provide. to other edge nodes traversing NAT.
Recognizing those we can determine features that have to
be included in a communication network for future work. A Table I points out, that tinc and cloudVPN meet the re-
matrix matching the requirements of agent-based systems with quirements in connectivity and security. They provide routing
P2P capabilities is provided in Table I. mechanisms to deal with nodes that disconnect from the over-
A major advantage of P2P is the ability to traverse NAT and lay and implement NAT traversal mechanisms to set up direct
Firewalls. P2P networks are approved to deal with flexible communication links between hosts. Although they initially
 5

TABLE I
C OMPARISON OF DECENTRALIZED COMMUNICATION NETWORKS FOR S MART G RID

Bootstrapping Security Connectivity Self-Healing Storage and search of

information
Mesh-VPN
Tinc and cloudVPN The traffic between Any-to-any connectivity Automatic routing It is not intended to
• tinc require pre-configured network nodes is is supported by rout- mechanisms will have extra
• cloudVPN connections. encrypted by using ing traffic in the overlay route traffic across applications besides
SSL/TLS. network. Both tinc and other nodes if one of the virtual private
Nodes only accept cloudVPN provide NAT the nodes becomes network.
connections from traversal methods to set unavailable.
authorized nodes by up direct links between
using public and nodes.
private keys.
Edge nodes have to N2N uses symmetric Any-to-any connectivity
• N2N connect to a known encryption (Twofish) is given by packet
supernode first. between nodes. forwarding in the
A differentiation by overlay on layer 2.
user or single user re- Direct communication
vocation is not pos- between devices is
sible because every supported by traversing
node uses the same firewalls and NAT.
key.
For Hamachi Hamachi uses Any-to-any connectivity
• Hamachi networks there is SSL/TLS for by routing in the
a dedicated cloud encryption. overlay. Dynamic direct
service available that Authentication is links are supported by
every network client handled with public NAT traversal.
uses to find other and private keys.
nodes. Every client is
keeping a list of
authorized nodes.
P2PVPN is using a SSL/TLS encryption Any-to-any connectivity
• P2PVPN BitTorrent tracker for Optional hierarchical by routing in the over-
initially retrieving a PKI for Authentica- lay. Direct links are es-
list with active net- tion allowing users to tablished if peers are
work nodes. invite other users public. NAT is not sup-
ported.
Peer-to-Peer networks
Gnutella uses GWe- Several clients Gnutella supports the Gnutella is a highly Gnutella supports
• Gnutella bCache for caching provide an encryption dynamic setup of direct self-organizing and distributed search
and retrieving a num- of the traffic between connections between self-healing network routines based upon
ber of active Gnutella two Gnutella nodes. peers for file exchange that can handle high DHT.
on/from a webserver. Since Gnutella has using NAT traversal. fluctuations of nodes.
been designed to Relaying in the overlay
be an open network is only implemented
no authentication is for search and ping
supported. messages.
BitTorrent clients use Initially BitTorrent BitTorrent does not Since there is no Several BitTorrent
• BitTorrent a tracker to retrieve did not support any support relaying overlay network for clients provide
a list of active P2P security features. messages in the overlay. relaying messages distributed search
nodes. Later versions include It does provide direct BitTorrent does not methods based upon
encryption and user P2P connections using provide self-healing DHT to achieve a
authentication to keep NAT traversal mechanisms. fully-decentralized
file-transfers private. trackerless network.

were considered as suitable communication networks they with low capability. The automatic management of the overlay
suffer in bootstrapping. Every connection has to be configured can deal with appearing and disappearing nodes. Unfortunately
manually. This will not work out for an agent-based system, it does not provide the ability to exclude a single network node
where including a new agent would require to adjust every without changing the symmetric key on every other node and
other agents configuration. Hence, more dynamic systems have therefor lacks in user management.
been considered, such as Hamachi, N2N and P2PVPN. The idea of P2PVPN in general matches the requirements
Hamachi is very suitable according to its self-organization and of agent-based systems especially concerning bootstrapping.
connectivity features. New nodes can easily be added to the It makes use of BitTorrent trackers and supports peer-
network without manual configuration. The bootstrapping un- information exchange if at least one node is known. Direct
fortunately uses the companies server application and Hamachi communication links are created between two nodes if it is
therefore cannot be considered as an independent system. possible. If not, it communicates through other nodes which
N2N in our evaluation scores in terms of efficiency as the act as a relay unit. Still it is not a sufficient communication
application focuses on a very tiny implementation for devices network for agents in heterogeneous networks as it lacks in
 6

NAT traversal and scalability. [7] R. Becker, “Entwurf und Betrieb von Regelungs- und Monitoringsys-
temen für dezentrale Energiesysteme auf der Basis von verteilten
Embedded Systems,” Ph.D. dissertation, Universität Kassel, 2008.
[Online]. Available: http://kobra.bibliothek.uni-kassel.de/bitstream/urn:
VI. C ONCLUSION nbn:de:hebis:34-2009061828307/3/DissertationRainerBecker.pdf
[8] M. Rossberg and G. Schaefer, “A survey on automatic configuration
In this paper we described the need for decentralized of virtual private networks,” Computer Networks, vol. 55, no. 8, pp.
communication in the Smart Grid especially for agent systems. 1684 – 1699, 2011. [Online]. Available: http://www.sciencedirect.com/
science/article/pii/S1389128611000053
The key functionalities of decentralized P2P communication [9] D. Wolinsky, K. Lee, P. Boykin, and R. Figueiredo, “On the design
networks have been pointed out. P2P networks demonstrate of autonomic, decentralized vpns,” in Collaborative Computing: Net-
how robust decentralized networks should be built. They allow working, Applications and Worksharing (CollaborateCom), 2010 6th
International Conference on, oct. 2010, pp. 1 –10.
fast communication with direct communication links between [10] Michele and Amoretti, “Towards a peer-to-peer hydrogen economy
nodes and demonstrate the handling of high node fluctuations. framework,” International Journal of Hydrogen Energy, vol. 36, no. 11,
These key functionalities should be implemented in mesh- pp. 6376 – 6386, 2011. [Online]. Available: http://www.sciencedirect.
com/science/article/pii/S0360319911006288
VPN to support an agent-based Smart Grid in the future. [11] A. Sanna, C. Zunino, and L. Ciminiera, “A distributed jxta-
Agent-based systems will require a middle layer offering basic based architecture for searching and retrieving solar data,” Future
connectivity between applications. The underlying network Generation Computer Systems, vol. 21, no. 3, pp. 349 – 359,
2005, p2P computing and interaction with grids. [Online]. Available:
must help with finding agents, as well as storing and searching http://www.sciencedirect.com/science/article/pii/S0167739X04000561
for information. They therefore have to adopt methods from [12] G. Deconinck, W. Labeeuw, S. Vandael, H. Beitollahi, K. De Craemer,
P2P networks. As pure P2P networks limit the possibilities and R. Duan, Z. Qui, P. Ramaswamy, B. Vande Meerssche, I. Vervenne, and
R. Belmans, “Communication overlays and agents for dependable smart
flexibility of the actual application, we conclude autonomic power grids,” in Critical Infrastructure (CRIS), 2010 5th International
mesh-VPN are the most suitable approach to create an ap- Conference on, sept. 2010, pp. 1 –7.
plication independent plug-and-play communication network [13] H. Beitollahi and G. Deconinck, “Peer-to-peer networks applied to power
grid,” Proceedings of the International conference on Risks and Security
in shared heterogeneous networks. Mesh-VPNs implement of Internet and Systems (CRiSIS), p. 8, 2007. [Online]. Available:
NAT traversal and self-organization from P2P networks and http://www.esat.kuleuven.be/electa/publications/fulltexts/pub 1685.pdf
hence are more flexible regarding the encapsulated application [14] I. Timmermans and G. Sliepen, tinc Manual, 2011. [Online]. Available:
http://www.tinc-vpn.org/documentation-1.1/tinc.pdf
protocol. Several mesh-VPNs have been considered for an [15] W. Ginolas, “Aufbau eines virtuellen privaten Netzes mit Peer-to-Peer-
agent-based Smart Grid, but it turns out more effort has Technologie,” 2009. [Online]. Available: p2pvpn.org/P2PVPN-45.pdf
to be invested on bootstrapping, user management and the [16] LogMeIn, LogMeIn Hamachi User Guide, 2012. [Online].
Available: http://help.logmein.com/KnowledgeLinkPublicRenderer?id=
overlay topology and application services such as searching kA230000000LnQ0CAK
for near-by agents. Addressing those issues and improving the [17] L. Deri and R. Andrews, “N2n: A layer two peer-to-peer vpn,” 2008.
state of the art mesh-VPNs would enable a variety of new [Online]. Available: http://luca.ntop.org/n2n.pdf
decentralized applications in the Smart Grid, not just agent-
based systems. VIII. B IOGRAPHIES

Simon Fey graduated in electrical engineering and

VII. R EFERENCES information technology at Offenburg University of
Applied Science. He worked at Fraunhofer Institute
[1] G. Rohbogner, S. Fey, U. Hahnel, P. Benoit, and B. Wille-Haussmann, for Solar Energy Systems (ISE) on acquisition and
“What the term agent stands for in the smart grid. definition of storage of measurement data from power plants.
agents and mulitagent systems froma an engineer’s perspective,” in After several months as research assistant at the ISE
Computer Science and Information Systems (FedCSIS), 2012 Federated he gained a scholarship and is now member of the
Conference, 2012. graduate school Small Scale Renewable Energy Sys-
[2] National Institute of Standards and Technology (U.S.). Office of the tems under the supervision of Prof. Dr. Andreas H.
National Coordinator for Smart Grid Interoperability, NIST framework Christ. He is currently writing his thesis on IT-based
and roadmap for Smart Grid interoperability standards, release 1.0. communication structures and system architectures
U.S. Dept. of Commerce, National Institute of Standards and Technol- for renewable energy systems.
ogy, Office of the National Coordinator for Smart Grid Interoperability,
[Gaithersburg., MD], 2010.
[3] GridWise Architecture Council, “Interoperability context-setting
framework,” 2008. [Online]. Available: http://www.gridwiseac.org/
pdfs/interopframework v1 1.pdf Pascal Benoit graduated in electrical engineering
[4] T. Logenthiran, D. Srinivasan, A. Khambadkone, and H. N. Aung, and information engineering at Karlsruhe Institute
“Multiagent system for real-time operation of a microgrid in real-time of Technology (KIT). He specialized in renewable
digital simulator,” Smart Grid, IEEE Transactions on, vol. 3, no. 2, energies and information technologies. He is work-
pp. 925 –933, june 2012. [Online]. Available: http://ieeexplore.ieee.org/ ing at Fraunhofer Institute for Solar Energy Systems
stamp/stamp.jsp?tp=&arnumber=6180026&isnumber=6204228 (ISE) in the department of Smart Grids. There he is
[5] T. Linnenberg, I. Wior, S. Schreiber, and A. Fay, “A market-based multi- active in the field of smart integration of electric ve-
agent-system for decentralized power and grid control,” in Emerging hicles, information an communication technologies
Technologies Factory Automation (ETFA), 2011 IEEE 16th Conference for energy systems and control systems of large-
on, sept. 2011, pp. 1 –8. [Online]. Available: http://ieeexplore.ieee.org/ scale concentrator photovoltaic (CPV) plants.
stamp/stamp.jsp?tp=&arnumber=6059126&isnumber=6058966
[6] Cisco Systems Inc. (2012, May) Cisco visual networking index:
Forecast and methodology, 2011-2016. Online. [Online]. Available:
http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/
ns705/ns827/white paper c11-481360 ns827 Networking Solutions
White Paper.html
 7

Gregor Rohbogner received the Dipl.-Wirt.-Ing.

degree from the RWTH Aachen University in Juli
2011. Since 2010 he is a researcher within the
Fraunhofer Institute for Solar Energy Systems (ISE).
First he focused on topics of in-house and Smart
Grid communication for his diploma thesis. Since
mid of 2011 he works as research fellow on multi-
agent systems for Smart Grid applications. Further
interest include Peer-to-Peer networks, Smart Home
technologies, optimization of decentralized energy
resources and network control technologies.

Andreas H. Christ received the Doktor degree in

electrical engineering from Technische Universität
Darmstadt, Germany, in 1988.
He joined the Central Research and Development
Laboratories and later on the Public Communica-
tion Department of Siemens AG, Munich, Germany,
where he was engaged in microwaves, mobile com-
munications and fiber-to-the-home projects.
In 1993, he became a Professor at Offenburg Uni-
versity of Applied Sciences, Germany. In 1997,
he became Head of the Department ’Media and
Information’ and since 2007 he is Vice-President of Offenburg University of
Applied Sciences. Also, as Head of the Information Centre he is responsible
for the deployment of and support for e-learning and electronic media for
education and research purpose.
His professional interests include technologies for mobile Internet.

Christof Wittwer was born in 1967 and stud-

ied electrical engineering at the University Kaiser-
slautern. He has been working at the Fraunhofer
Institute for Solar Energy Systems ISE since 1992.
Between 1996 and 1999 he worked as a research
assistant in the fields of building services engineer-
ing and systems control receiving his promotion in
1999 at the University of Karlsruhe. Since 1999
Christof Wittwer has been working as a project man-
ager at Fraunhofer ISE concentrating on research
and development of networked control systems for
decentralized power generation systems. Since 2010 he is head of Department
Smart Grids at Fraunhofer ISE.