A Multi-Level Communication Architecture of Smart Grid Based on Congestion Aware Wireless Mesh Network

*Yichi Zhang, Weiqing Sun, *Lingfeng Wang, Hong Wang, *Robert C. Green II, and *Mansoor Alam
*Department of Electrical Engineering and Computer Science Department of Engineering Technology University of Toledo 2801 W. Bancroft St. Toledo, OH 43606 Email: Lingfeng.Wang@utoledo.edu
Abstract The advent of the smart grid promises to usher in an era that will bring intelligence, efficiency, and optimality to the power grid. Most of these changes will occur as an Internet-like communications network is superimposed on top of the current power grid. The communication environment of the smart grid should be more robust, reliable, and efficient. In order to address this issue, this work proposes a Smart Grid Communication Network (SGCN) by deploying wireless mesh network technologies consisting of 802.15.4 Zigbee, 802.11, and the WiMAX standards. These are applied in different levels of the home area network (HAN), neighborhood area network (NAN), and the Wide Area Network (WAN). Simulation results demonstrate that this is a promising methodology for a more efficient communication environment.

WMN

Wireless Mesh Network I. INTRODUCTION

Nomenclature: AMI CAC E&S C EDS HAN MCS MM NAN SCADA C SGCN SM SMDC WAN Advanced Metering Infrastructure Central Access Controller Energy and Service Corporations Energy Distribution System Home Area Networks Meter Controlling System Metering Module Neighborhood Area Networks Supervisory Control And Data Acquisition Controller Smart Grid Communication Network Service Module Smart Meter Data Collector Wide Area Networks

The advent of the smart grid provides the world with an efficient and intelligent approach of managing energy supply and consumption. Consumers and energy suppliers alike can take advantage of the convenience, reliability, and energy savings provided through real time energy management. One of the great advantages of the smart grid is the installation of a completely new, two-way communication network between energy suppliers and their customers. This allows the smart grid to be considered as a power grid that has an integrated data communication network providing the collection and analysis of data at all levels in real time. The real-time communication ability of the smart grid will enable utilities to optimize and modernize the power grid in order to realize its full potential [1]. This communication network will provide a number of new energy concepts including real-time metering and pricing, intelligent load shedding, consumption management, cost savings from peak load reduction and energy efficiency, integration of plug-in hybrid electric vehicles for grid energy storage, and the integration of alternative and distributed generation sources including photovoltaic systems and wind turbines. This new communication network will be constructed using various communication paths including fiber optic cable, twisted pair, broadband over power line, and wireless technologies [2]. In order to enable these new technologies, the advanced communication network mechanisms of the smart grid will bring new conflicts and requirements. For instance, a typical computational network that is applied to the smart grid may not satisfy the ability to control power among the data communications network. Also, the frequent communications between the components in the smart grid may increase the power consumption and congestion in the whole network. Network scheduling issues such as channel assignment and

routing efficiency should also be considered to ensure the fluent transmission of data [3] while avoiding high rates of congestion. The application of standard topologies and protocols should also be considered to fulfill the complicated requirements of the smart grid architecture. Due to these issues, an integrated strategy to provide a robust communication network for the smart grid is extremely important. In this paper a new architecture for a multi-level communication network named the smart grid communication network (SGCN) using a wireless mesh network environment is proposed. This communication network allows the integration of applicable components in the smart grid and also allows the appropriate communication scenarios among this network. A mesh network also provides a reliable communication environment with high speed communication and power efficient routing. The remainder of this paper is organized as follows: Section II provides a literature review of this research field, Section III presents the architecture and design of the SGCN that is proposed including the topology selection, protocol selection and the routing design, Section IV details the simulation of the communication scenario and analysis of the SGCN in the wireless mesh network environment, and Section V provides conclusions and future work. II. LITERATURE REVIEW

prohibitive or impractical in wireless mesh networks [9]. WiMAX may also provide lower costs and higher performance in the wireless mesh network [10]. Another aspect that affects the efficient communication of components in the SGCN is network routing. In [3], [11], and [12] the constraints of the wireless mesh network are analyzed and the optimal communication routing is explored through the use of the primal-dual algorithm. In part III of this paper, this algorithm is applied to find the optimal routing for communications between the components of the SGCN. III. SYSTEM DESIGN

The communication architecture of the smart grid is quickly becoming an important issue due to the growing research related to the smart grid. The most popular issue being studied is the advanced metering infrastructure (AMI) and its network architecture [1], [4], [5]. An AMI is viewed as a fundamental technology for the smart grid. It is composed of four parts (A smart meter, customer gateway, AMI communication network, and headend) and can generally be considered as a meter providing two-way communications, automated meter data collection, outage management, dynamic rate structures, and demand response for load control [4]. An architecture for AMI that is composed of several separate communication networks is illustrated in [5], but this design limits the ability to control the network as a single system. The SGCN is designed to alleviate this issue as there are three layers that may be controlled in both a centralized or de-centralized manner. In the SGCN it is crucial to choose an efficient communication topology and standard protocols. In light of this, many works about network topologies and standards were reviewed. In [2] and [6] the wireless mesh network topology was discussed and compared with other topologies. It was concluded that the mesh network could supply a reliable, robust, and cost-effective topology for communication. In [7] the construction and parameters of wireless mesh networks in different environments are discussed. The standards which are suitable for the wireless mesh topology are also studied. For instance, Zigbee is a lowcost and low-power network standard which can be widely deployed and utilized in the wireless mesh network environment [8]. The 802.11n Wi-Fi standard can also enable a range of applications that were previously cost

A. Smart Grid Communication Network Architecture When considering the large communication network that will exist in the smart grid, a three-layer network composed of home area networks (HAN), neighborhood area networks (NAN), and wide area networks (WAN) may be assumed as a reasonable infrastructure design. Each network layer owns several modules or controlling systems which may be extended to include new modules and systems in the future. The first layer of the network is the home area network (HAN) which consists of the service module (SM), the metering module (MM), and the meter controlling system (MCS). The SM will provide real-time energy cost and consumption data to the consumers while the MM will record the consumption of the energy in a consumers home. The MCS will collect and control the information transmitted from SM and MM. The second layer is the neighborhood area network (NAN). The NAN is a large metering network which is formed by connecting multiple MCSs of HANs that are geographically near each other. The NAN will consist of two additional components: The central access controller (CAC) and the smart meter data collector (SMDC). The CAC can be considered as the interface that manages the communication between HANs and the energy supplier or utility. The SMDC will be a wireless node that is in charge of the metering records of the whole community. The final layer is the WAN. The WAN provides broadband wireless communication between the NAN, substations, other distributed grid devices, and the utility. This layer will consist of the energy distribution system (EDS), the supervisory control and data acquisition (SCADA) controller, and the energy and service corporations (E&SC). The EDS will remain in charge of the energy and metering data distribution. The SCADA controller provides the utility or grid operator with distributed process control in order to manage the distribution grid elements. The collected metering and control information is then transmitted to the E&SC which is able to make a price judgment. A graphical depiction of all three levels of the proposed network can be seen in Fig. 1. The controllers of each layer (MCS, CAC, SMDC, and SCADA) own their controlling modules. Controlling modules of the upper layers can also control the modules of the connected lower layer. For instance, the upper module can send commands to the lower modules for checking the

metering values so that the multi-level and distributed control can be achieved. This satisfies the centralized and distributed controlling requirement of the smart grid.

requirements differ across the different layers. In the lower HAN and NAN layers the requirements include high network capacity and data transmission. In the third layer stability and reliability of the communication network is the main requirement. In the HAN, the network standard chosen is 802.15.4 Zigbee. This protocol provides radio communication specifically suitable for personal and home area networks [8]. It has three types of available bands: 868-868.6 MHz with 1 channel for European utilities, 902-928 MHz with 10 channels for North American utilities, and 2.4 GHz with 16 nonoverlapping channels for worldwide utilities. The 2.4 GHz band is chosen since it supports the data rate of 250 kbps as over-the-air data transmission, which is higher than the data rates of other two bands. Also, as it has 16 channels. The maximum wireless capacity is calculated in (1) where Cmax represents the maximum wireless capacity, r represents the data rate, and N C means the non-overlapping channels. Thus the maximum capacity of one data link under the 802.15.4 Zigbee standard is 4Mbps.

Figure 1. Three layer architecture of the SGCN

Cmax = r N C

(1)

The SGCN is a general communication architecture for the smart grid which can support a wide variety of applications. Modules can be added into each layer by different utilities to serve various purposes. For instance, intrusion detection system (IDS) modules could be plugged into each layer to monitor the activities between each component and record and identify suspicious communications. This would allow the cyber security of the network to be improved. B. Network Topology and Standards Selection Because of the recent implementation of standards-based wireless communication protocols such as IEEE 802.15.4 Zigbee, the widespread use of 802.11 protocols, the use of 802.16 WiMAX, and the need for distributed sensing technology, it can be assumed that large parts of each network layer will exist in the form of a wireless mesh network [2]. The communication network topology chosen for the SGCN architecture is the wireless mesh network (WMN). The mesh network topology is adopted and is beneficial due to its redundant communication paths which compensate for natural failure as well as its scalable, dynamic and self-healing capacities [2]. Mesh networks also provide multiple, redundant communication paths that can compensate for the failures brought on by the break-down of communication between nodes due to failures, congestion, and other maladies. It is known that the majority of utilities have applied the wireless mesh system for smart grid components. The earth2techs Katie Fehrenbacher estimated that 96% of the market is owned by wireless mesh networks when smart grid network installations are considered [13]. Since thousands of network standards are able to support the smart grid protocols and networks, the requirements should be set for selecting the optimal network standards for the communication network of the smart grid, though these

In the NAN, 802.11n is chosen as the communication standard. Compared with other Wi-Fi standards, 802.11n increases the over-the-air data rates to a maximum of 300 Mbps on the 2.4 GHz band. This band is also shared with 802.15.4 Zigbee in the PHY layer [14]. The number of possible un-overlapping channels achieved is 23. The communication range is the largest when compared with the other 802.11 standards as the total range can be up to 250 m outdoors. By using (1), the maximum wireless capacity of links under the 802.11n standard is 6,900Mbps. In the WAN, since the elements are mostly fixed and physically distant from each other, both wireless and wired networks will be applied to assist the availability and reliability between the communications of the smart grid elements. In our network, the large amount of data that is transmitted is sent via wired communication techniques such as power line communication (PLC), while the control information is transmitted by a wireless network. The wireless standard chosen is the WiMAX 802.16-2009. This protocol is currently utilized with a 70Mbps data rate under the spectrally efficient orthogonal frequency division multiplexing (OFDM) mechanism. The most crucial reason for choosing WiMAX is that it can provide communication at a distance as large as 2 miles. In this paper, only the wireless portion of the network is simulated for simplicity [10]. By applying the WMN topology and maintaining high data rate standards, the SGCN can provide a robust communication environment that is also congestion aware. For instance, when one access point of a smart meter is broken or flooded with traffic, the user can still read their metering information by accessing data from the other communication nodes or components. Also, by applying the broadband and high capacity wireless standards, SGCN is able to satisfy the requirement of transmitting a large amount of data.

C. Network Constraints and Optimization Here we use a directed graph, G = (V , E ) , to illustrate the multi-hop multi-channel network [3]. Each node V c can be considered as an AMI component for measuring electricity consumption and managing communications with other nodes or higher layer components. Each edge, E , represents the wireless links between the nodes. In this graph interference amidst communications is not considered for simplicity. Only C (e) C channels can be activated on any link e of the network where each channel is orthogonal to another. The data sent on each channel is transmitted by one and only one radio, i . Each channel of the data link has a capacity of ci ( e) . This is a constant showing the maximum data flow on that channel. The information flow currently transmitting on channel i of link e is denoted as f i ( e) . In order to make sure that the data flow can be transmitted through the channels smoothly, the constraints for the link or channel and all flows should be set appropriately. Suppose the t t scheduling variable xi ( e) exists as xi ( e) = 1 when data is transmitting on link channel i of link e during time slot t . Thus, the maximum number of the active channels, C ( e ) , during time slot t is:

needed in the network. Because of the constraints above, it is required that SF should not be smaller than 1. The primal dual algorithm is described in Fig. 2.

While w(j) < 1

j =1

w( j ) = , j {1, 2,..., } b=0

For m = 1,2 ,...,M r = r(m) While r > 0 Set w(j) =

(S
j S i

(j)
j

ci(e)

Pmin (s(m),d(m) ) = min

m M

w(j)

x ( e ) C ( e)
t i iC

e E

(2)

= min( f ( Pmin )) = min{ r , u} r = r f i (e ) = f i (e ) +

w( j ) = w( j ) (1 + endWhile endFor b = b +1 endWhile f i(e) g = max c (e) j S i b SF = g
Figure 2. Primal Dual algorithm [3]

f ( Pmin )

Also, there are sets of link and channel constraining pairs ( S1 , S2 ,..., S ) between each pair of nodes which are allowed to communicate with each other directly. In each pair, S j means the constraining set belongs to channel i of link

e . Each pair has a RHS constant, ( S j ) , according to the

following constraint

1 f i ( e) 1 , j {1, 2,..., } ( S j ) ( e,i )S j ci (e)

(3)

M source destination pairs denoted as ( s (m), d (m)) . In m each pair the flow allowed on channel i is set to be pi ( e) and the total flow, r ( m ) , transmitted between the pair can be m calculated as pi ( e) = r ( m ) according to the
e = s ( m ) iC

The nodes communicating with each other consist of

In this algorithm w( j ) is the weight of the constraining pair of link i channel

e and ( j ) is the weight of each set

S j whose value is calculated by the distance between two

nodes. When searching for the optimal route, all the routes starting from the source node and ending with the destination node will be counted and the accumulation of the constraining pairs weights will be calculated. Then the route which has the lowest value of weight accumulations will be labeled as the shortest path. The desired data flow, r ( m ) , is distributed to the corresponding node, and this source destination pair, which is also called a commodity pair, can find the optimal route with the shortest distances and link weights. By using this algorithm, the problem of finding the optimal path and the

constraint

( e) 1 mM c ( e) 1 S j S j i

m i

(4)

The primal-dual algorithm is applied to find the optimal path for communication flows. Set SF SF as the maximum scaling factor where its value denotes the total slack capacity

flow distribution is settled. Through the application of the primal-dual algorithm to find the shortest path of communication, a communication network could be provided to satisfy the requirement of efficient power consumption in the smart grid by the SGCN. IV. SIMULATION AND ANALYSIS

A. Grid Topology The wireless mesh communication scenario that uses SGCN is simulated using MATLAB as it is a powerful tool for simulating a complex communication system. Fig. 3 shows the simulation of the multi-level, three layer smart grid communication network that has been proposed in Fig.1. In Fig. 3, a mesh network with 20 HAN nodes which are scattered randomly in a 500 500 square is generated. These nodes representing one of the SM, MM, and MCS compose a single HAN. According to the rules of the communications among the nodes in a wireless mesh network, a node can directly communicate with another node by the link between them as long as it is located less than a specified distance away. Here the direct communication distance as set to 100m since the longest communication distance of Zigbee is 150m. Remote node pairs that need to communicate with each other will use the multiple-hop approach by sending their communications through other nodes. The medium sized nodes (21 to 23) represent the CAC or SMDC components in each NAN. These components can directly communicate with the nearest nodes in HAN by using the 802.11n standard. Remote nodes in HAN that need to communicate with other nodes in the NAN can send the information through other HAN nodes. Since one NAN node may route the metering and detection information from multiple HAN nodes, the capacity of the each node and link in the NAN is set higher.
500
13 16

communication is consumed at the intermediate nodes. As such, it is necessary to make sure that there is enough power available at these intermediate nodes. Since only the wireless communication is simulated, the data transmission among the components in the WAN by using wired network is omitted and only the controller in the WAN is simulated. The HAN nodes do not have communications with the WAN node in the simulation since the metering information transmitted to the WAN node is also omitted. The wireless communication standard between a NAN node and the WAN node uses the WiMAX protocol. Each link between a WAN and HAN has only one channel in this study. The data rate and maximum capacity of the links are also listed in table I.
TABLE I. Network HAN HAN to NAN NAN NAN to WAN WAN DATA RATE AND LINK CAPACITY OF NETWORK Wireless Communication Parameters
Data rate(bps) Max Capacity(bps)

250k 300M 300M 70M 70M

4M 6900M 6900M 70M 70M

450
4 8 12 7 11 19 3 15 20

400

23

Network width/m

350 300 250 200 150

22

24

10

18

B. Communication Routing Optimization and Analysis The source and destination node pairs are randomly picked from HAN nodes and NAN nodes. The source nodes are tasked to send out packets to the destination nodes directly or by using multiple hops. In this simulation, five commodity pairs are set with each pair labeled using one color. The corresponding route is also labeled with the same color that is used by the commodity pair. If the node is used as the source and the destination, its color would be overlapped. The shortest routing for the commodity pair is found by using the primal dual algorithm described in Section III-C. In order to more realistically simulate the data transmission, the number of channels and the data rate is randomly set on each link under the rule of the network constraints. This means that in the HAN a maximum 16 channels can be opened simultaneously on one link. In the NAN the highest number of channels is 23. Fig. 4 illustrates the communication routing between the commodity pairs and the optimal routings are recorded in table II. In table II, it can be seen that the route chosen by commodity pairs (16, 18) use node 15 and 19 for multiple hopping nodes. Although the distance between node 15 and 18 is shorter than the distance hopping from 15 to 19 and 19 to 18, the link between 15 and 18 has a higher weight value since fewer channels are open on this link. This is an example of how using only the shortest path may bring about communication congestion. Thus, by applying the Primal Dual algorithm, optimal routing of the communication could be achieved.

100 50 0
1

2 5 9 14 17

21

100

200

300

400

500

600

700

800

900

1000

Network length/m
Figure 3. Smart grid communication network topology

Adding the largest node (24) into the WAN completes the entire 1000 500 network. The node in a WAN can be considered as an EDS, a SCADA Controller, or the E&SC. As the communication between the WAN node and NAN node is completed across a larger distance, the majority of power for

500
13 16

450
4 8 12 7 11 19 3 15 20

architecture and more intelligence will be embedded into the modules to establish a multi-agent based communication architecture for smart grid applications.
23

400

Network width/m

350 300 250 200 150

REFERENCES
[1]
22

24

[2]

10

18

[3]
21
5 9 14 17

100 50 0
1

[4]
400 600 800 1000

200

[5]

Network length/m
Figure 4. Optimal communication routing selection in the SGCN [6] [7]

TABLE II.
Source node

THE ROUTING TABLE OF THE COMMUNICATION AMONG NODES

Intermediate nodes

Destination Node

3 7 11 16 20 15 11 15 10 19 5 11 15

10 23 15 18 1

[8]

[9] [10] [11]

V.

CONCLUSION AND FUTURE WORK

[12]

This paper proposed a new communication network architecture, the SGCN, for the smart grid. This architecture provides a multi-level, robust, and reliable communication network using a wireless mesh network with high capacity wireless standards. It also supports optimal communication routing by using the primal-dual algorithm for smart grid communication. In future work, fault-tolerance capability of the network will be developed to improve the robustness of this

[13]

[14]

S. Keemink, B.Roos. (2008) Security analysis of Dutch smart metering systems, [Online]. Available: http://staff.science.uva.nl/~delaat/sne2007-2008/p33/report.pdf Daintree Networks. (2007). Whats so good about mesh networks? [Online]. Available: http://www.daintree.net/downloads/whitepapers/mesh-networking.pdf M. Kodialam and T. Nandagopal, Characterizing the Capacity Region in Multi-Radio Multi-Channel Wireless Mesh Networks, In Proc. ACM MobiCom, 2005. C. Bennett, and D. Highfill, Networking AMI Smart Meters, Energy 2030 Conference, 2008. ENERGY 2008. IEEE, 17-18 Nov. 2008 R. Berthier, W. Sanders, and H. Khurana, Intrusion Detection for Advanced Metering Infrastructures: Requirements and Architectural Directions, In Smart Grid Communications (SmartGridComm), 2010 First IEEE International Conference on, pp. 350355. IEEE. P. Yi, Y. Wu, F. Zou, and N. Liu, A Survey on Security in Wireless Mesh Networks, IETE Tech, pp.:6-14 G.Hernandez, C. F., Ibarguengoytia-Gonzalez, P. H. and Perez-Diaz, J. A. Wireless Sensor Networks and Applications: A Survey, IJCSNS International Journal of Computer Science and Network Security, 7(3), pp.264-273, 2007 S.Safaric, and K.Malaric,"ZigBee wireless standard," 48th International Symposium ELMAR-2006, pp.259-262, Zadar, Croatia, 07--09 June 2006. M. J. Lee and al.,"Emerging Standards for Wireless Mesh Technology," IEEE Wireless Communication, April 2006. H. Wei, S. Ganguly, A. Izmailov, and Z. Haas, "Interference-Aware IEEE 802.16 WiMax Mesh Networks,"in Proc. 61st IEEE VTC,2005. M.Kodialam, and T.Nandagopal, Characterizing Achievable Rates in Multi-hop Wireless Networks:The Joint Routing and Scheduling Problem, ACM Mobicom,September 2003. M.Kodialam, and T.Nandagopal, Characterizing Achievable Rates in Multi-hop Wireless Mesh Networks with Orthogonal Channels, To appear in IEEE/ACM Transactions of Networking, September 2005. Manoa Innovation Center, The Case for Using Wireless Mesh Technology in Smart Grid Communication Networks,, [Online]. Available: http://www.concentrissystems.com/downloads/SmartGridWMAdvantages.pdf A.Sikora and V.F.Groza, Coexistence of IEEE802.15.4 with other systems in the 2.4 GHz-ISM-Band, in Proc. IEEE Instrumentation & Measurement Technology Conference, Ottawa, May 2005, pp.17861791.