Y. Yan and W. Su, "A Fog Computing Solution for Advanced Metering Infrastructure", 2016 IEEE Power and Energy Society Transmission and Distribution Conference and Exposition, Dallas, TX, U.S.A. May 2-5, 2016.

A Fog Computing Solution for Advanced Metering Infrastructure Yu Yan

Wencong Su

Department of Electrical and Computer Engineering University of Michigan-Dearborn 4901 Evergreen Road, Dearborn, Michigan, USA [email protected]

Department of Electrical and Computer Engineering University of Michigan-Dearborn 4901 Evergreen Road, Dearborn, Michigan, USA [email protected]

Abstract— With rapid increase in deployment of advanced metering infrastructure, the amount of the data collected increases dramatically. Incorporation of such voluminous data requires wide-ranging transformation of the existing metering infrastructure. However, as the number of smart meters increases to more than hundreds of thousands, it would become increasingly clear that the state-of-the-art centralized information processing architecture will no longer be sustainable under such big data explosion. In this short communication letter, we propose an implantable data storageand-processing solution for improving the existing smart meter infrastructure. The practicality of the proposed solution is validated on a proof-of-concept testbed. Index Terms-- Fog Computing; Smart Meter; Smart Grid; Advanced Metering Infrastructure (AMI); and Big Data.

I.

INTRODUCTION

In North America, the number of smart meters installed has grown from 6% of households in 2008 to 89% in 2012, and by 2019, 30 million homes and small businesses will have smart meters [1]. For example, Austin Energy in Texas has implemented 50,000 smart meters, and these smart meters send data every fifteen minutes to the data center, which requires 200TB of storage capacity as well as significant recovery redundancy. Also, it takes more disk space to manage than just storing the information. If Austin energy moves from 15-minute to 5-minute interval data exchanging, their data storage needs will grow to 800TB [2]. Smart meters support high sampling rate for data collection via a two way communication, while conventional electricity meters record data only once a month [3]. The massive volume of real-time data collected by smart meters will help the grid operators gain a better understanding of a large-scale and highly dynamic power system [4]. How to store and process the huge amount of data becomes a critical issue. When it comes to the large data storage and processing area, the centralized data center with relational database is a common choice. All the smart meter

Corresponding Author: Wencong Su ([email protected])

data are aggregated and processed in a central location via two-way communication channel. More specifically, the legacy of smart meter infrastructure needs to be improved substantially, in terms of the following four important aspects: Inflexible for expansion: Expanding the existing database infrastructure is very complex and time consuming. It is almost impossible to enhance hardware performance by just improving the hardware, without incurring additional cost for the unit [5]. Inefficient: As there is no efficient way of processing raw data with the existing centralized method [6], it requires an efficient information extraction process. Unreliable: A single point of failure in the chain of operations may degrade the reliability of the whole power system and make the entire database unreliable. Such database contains critical information for day-to-day system operations [7]. Moreover, High-frequency metering data which are required for efficient network operations may reveal customers’ private information [8]. High-cost: Some software companies (e.g., Oracle) have been investigating for new solutions to deal with the increasing amount of smart meter data. However, they focus mainly on creating a brand new system, which can end up as a rather costly investment proposal. As the number of smart meters increases to more than hundreds of thousands, it is rather intuitive that the current state-of-the-art centralized data processing architecture will no longer be sustainable under such big data explosion. Nowadays, a great amount of work ([5],[10]) focus on applying cloud-based solution for data-driven analytics in smart grid to deal with the data explosion issue. However, there are several main disadvantages of cloud computing. For example, cloud computing systems are internet-based and mainly depend on the internet connectivity. Smart meters send data package to the cloud every second, resulting in high demand of communication bandwidth. Data security and privacy are also critical issues when the sensitive smart meter data is aggregated in a cloud.

Y. Yan and W. Su, "A Fog Computing Solution for Advanced Metering Infrastructure", 2016 IEEE Power and Energy Society Transmission and Distribution Conference and Exposition, Dallas, TX, U.S.A. May 2-5, 2016. To address these challenges, Fog Computing, an extended II. FOG COMPUTING PLATFORM concept of cloud computing, provides more advantages on its For the proposed fog computing-like data storage-anddecreasing service delay and its dense geographical processing solution, smart meters are grouped to form a distribution. The small latency and proximity to end-users cluster. Each smart meter acts as a Datanode and one of the features lead to the fog paradigm well positioned for real time Datanodes functions as the Master node, which manages the big data and real time analytics [11]. The major differences file system. Master node stores the metadata that contains between cloud computing and fog computing is summarized information of file name and storage location. To maintain in Table 1 [12]. privacy, the metadata is stored only in readable form via a Table 1. The comparison between cloud computing and fog computing [12] Cloud Computing

Fog Computing

Latency

High

Low

Delay Jitter

High

Very Low

Location of Service

Within the Internet

At the Edge of Local Network

Distance between Client and Server

Multiple Hops

One Hop

Geo-distribution

Centralized

Distributed

Number of Computing Nodes

Few

Very Large

Location Awareness

No

Yes

Security

Undefined

Can be Defined

Attack on Data Enroute

High Probability

Very Low Probability

Support for Mobility

Limited

Supported

In addition to electricity data collection and power outage detection, smart meters have the potential to be nodes of a distributed computing architecture by fully utilizing the AMI computing power. Therefore, a cluster of idle smart meters can significantly reduce our reliance on the centralized data center and supercomputer, as illustrated in Figure 1.

Figure 1. The working principle of the proposed data storage and processing In this paper, we propose an implantable fog computinglike solution for enhancing the existing smart meter infrastructure in a reliable, cost effective and efficient manner. We also validate the practicality of the proposed solution on a proof-of-concept testbed, using a cluster of lowcost minicomputers. The concepts of the proposed fog computing solution can be extended to many applications, such as connected vehicles and cyber-physical systems.

specific decoding program. Also, Datanodes (i.e., smart meters) as the end-users can also reach the cluster if the Master node permitted. At every fixed time interval, data captured by the smart meter will be duplicated and split into small chunks (e.g., 64MB) and then distributed to the Datanodes inside the cluster. Figure 2 shows the working principle of the proposed data storage and processing scheme.

Figure 2. The working principle of the proposed data storage and processing Every smart meter will store a portion of the duplicated data chunks as back-up of its neighbor’s data, as a precautionary measure in case of the hardware failures. In addition, the Master node will continuously monitor the connection status of the Datanodes to ensure overall system reliability. As the nodes are not required to share any storage space with each other, additional nodes can be easily plugged-and-played as the data repository grows. Fog Computing is a paradigm that extends Cloud computing and services to the edge of the network, as illustrated in Figure 3. In the proposed AMI framework, each Smart Grid component (e.g., distributed renewable energy generator, smart home, and smart building) is attached to smart meter (i.e., Fog device). These Fog devices are interconnected with each other and linked to the Cloud. As smart meter is implemented at the edge of the smart grid network, the Fog computing provides low latency, location awareness, and improves quality-of-services (QoS) for streaming and real time applications [13].

Y. Yan and W. Su, "A Fog Computing Solution for Advanced Metering Infrastructure", 2016 IEEE Power and Energy Society Transmission and Distribution Conference and Exposition, Dallas, TX, U.S.A. May 2-5, 2016. We can control the NameNode and monitor the DataNodes on another PC via SSH (putty), and it is convenient to monitor all three nodes on the same screen. The data is appended in a table which contains the information about the ID, voltage, current, temperature and frequency. At each row, different categories of information are separated by a blank space. All the simulated smart meter data is stored in a ".txt" file in the Hadoop cluster with respect to the data format as described above. We use Hive to create a table at the NameNode side and load the data into the table to make the data ready for further processing using MapReduce. Once a table is created, the data format is organized in place ID: integer; Voltage: Real; Current: Real; Temperature: double; Frequency: double. In order to manage the power grid, the power grid manager needs to know where and when the abnormal frequencies and voltages occur at a constant time interval. In Figure 3. Cloud and Fog Computing in Advanced our case study, the objective is to select the rows whose Metering Infrastructure (AMI) frequency is higher than 60.2 or lower than 59.8. These values are regarded as the abnormal frequency. Then, we load III. CASE STUDY these specific rows of data into a new table, as shown in We set up a proof-of-concept demonstration using a Figure 5. Due to the page limit, we can only show the first cluster of low-cost single board PCs (e.g., Cubieboard), as ten rows of this table which contains the information about place ID, voltage, current, temperature and frequency. Then shown in Figure 4. we can perform more complex data mining techniques to better understand how power systems respond to any variations and fluctuation.

Figure 4. A proof-of-concept demonstration This cluster consists of 15 Datanodes, which can communicate with each other. Each Cubieboard is a Datanode and can also represent a smart meter. Moreover, we use a PC to act as the Master node to monitor this cluster. In the present case study, we preload all the raw data in a specific data format in each Datanode, which can maintain the privacy of the smart meter data. All the data can be managed only by a Hadoop MySQL-like language – Hive. Here, we simulate the process of the grid operator to find the abnormal frequency data monitored by these “smart meters”. Firstly, we create a table via the Master node, and then we load the frequency data monitored by the smart meter in the specific form into this table. Also, the table will be packaged, cut and distributed to Datanodes spontaneously. Then, we use Hive for searching this table and finding the abnormal frequency data to prevent power outage or analyze the root cause of the accident after a blackout. During the searching process, we plug in and out a certain node respectively to mimic real-world scenarios in power system operations.

Figure 5. The first 10 rows of the selected dataset The experiments demonstrate the following four features of the proposed solution. Plug-and-Play: The distributed data cluster using fog computing can automatically manage addition or removal of any Datanode without reconfiguring the entire operation system. Efficient: The self-loop processing procedure is based mainly on parallel computation, which is particularly efficient for large volume of data. Figure 6 shows the comparison between average processing times by the centralized (Ubuntu server, 3.6GHZ, 8GB memory) and fog computing-like (a cluster of low-cost distributed Datanodes) approaches. The superiority of the distributed solution stands out obviously when we increase the data size from 39.4MB to 10GB. The fast and intelligent processing capability of large volumes of data will facilitate smart grid operations [14].

Y. Yan and W. Su, "A Fog Computing Solution for Advanced Metering Infrastructure", 2016 IEEE Power and Energy Society Transmission and Distribution Conference and Exposition, Dallas, TX, U.S.A. May 2-5, 2016. ACKNOWLEDGMENT This work was supported by the New Faculty Startup Fund at University of Michigan-Dearborn and the National Science Foundation under Award EEC-0812121. The authors gratefully acknowledge the technical contributions of Ni Zhang and Shengyao Xu from the University of Michigan-Dearborn. REFERENCES [1] [2]

Figure 6. A Comparison on average data processing time [3]

Robust: Even after shutting down several Datanodes during the experiment, the entire data is still readable and writeable any time. During the turn-off transient and turn-off transient, the whole system is still in stable operation. Thus, if any of the Datanodes is broken, unlike the centralized system with backup storage, it need not be repaired or replaced immediately, which further reduces the soft cost associated with the existing smart meter infrastructure. The low-cost hardware supports both wired and wireless communication, including network cable, Wi-Fi, and Bluetooth, etc. Our testbed is based on the wired communication. Since that split data in each smart meter is unreadable, the different communication methods will not lower the security level of the system. Low-cost: When facing big data exposition, under today’s centralized architecture, the cost of the relational database included in the purchase of software with its license and the maintenance cost of the database center keeps rising all the time [15]. Furthermore, the authors in [16] mentioned that it is difficult and costly to expand the existing data centers. Our proposed fog computing-like framework is able to provide a cost-effective alternative to the expensive centralized data center, in terms of portability and low-cost hardware. The existing and emerging advanced metering infrastructure can be easily extended and tailored to adopt the proposed fog computing-like solution. IV.

[4]

[5]

[6] [7]

[8]

[9]

[10]

[11]

[12]

[13]

CONCLUSION

This paper proposes a fog computing-like solution for the existing smart meter infrastructure. The proposed solution can be easily incorporated into the existing smart meter infrastructure, at a little extra capital cost. In our future research, our aim would be to fully integrate the sensor, data storage, data processing devices and distributed controllers into a single smart meter. A user-friendly graphic user interface will be developed to visualize and archive the collected data and process commands and additional information. In addition, we will customize the MapReduce schemes in Hadoop infrastructure to meet the specific application requirements. As discussed earlier, the fog computing solution for AMI has some strength. However, it is worth to mention that it is still a long way to move the concept of fog computing to the real-world AMI at scale.

[14]

[15] [16]

S. Davies, “Smart Meters”, Engineering & Technology, vol. 7, pp. 4849, July 2012. H. Mc and E. Stanley, “Grid Analytics: How Much Data Do You Really Need,” in Proc. of 2013 IEEE Rural Electric Power Conference (REPC), Apr. 2013. J. Zhou, R. Q. Hu, and Q. Yi, “Scalable Distributed Communication Architectures to Support Advanced Metering Infrastructure in Smart Grid,” IEEE Trans. Parallel and Distributed Systems, vol. 23, pp. 1632-1642, Sep. 2012. W. Su, J. Wang, and D. Ton, “Smart Grid Impact on Operation and Planning of Electric Energy Systems”, Handbook of Clean Energy Systems, Edited by A. Conejo and J. Yan, Wiley, July 2015. R. C. Green, L. F. Wang, and M. Alam, “Applications and Trends of High Performance Computing for Electric Power Systems: Focusing on Smart Grid”, IEEE Trans. Smart Grid, pp.922-931, Jan. 2013. E. Santacana, G. Rackliffe, L. Tang, and X. Feng, “Getting Smart,” IEEE Power and Energy Magazine, vol. 8, pp. 41-48, Mar. 2010. X. Fang, S. Misr, G. Xue, and D. Yang, “Smart Grid – The New and Improved Power Grid: A Survey,” IEEE Communications Surveys & Tutorials, vol. 14, pp. 6-9, Dec. 2012. C. Efthymiou, and G. Kalogridis, “Smart Grid Privacy via Anonymization of Smart Metering Data,” in Proc. of 2010 IEEE International Conference on Smart Grid Communications, pp. 238-243, Oct. 2010. N. Zhang, Y. Yan, S. Xu, and W. Su, “A Distributed Data Storage and Processing Framework for Next-Generation Residential Distribution Systems”, Electric Power Systems Research, vol.116, pp.174–181, November 2014 Y. Simmhan, S. Aman, A. Kumbhare, and L. Rongyang, “Cloud-Based Software Platform for Big Data Analytics in Smart Grids,” Computing in Science & Engineering, pp. 38-47, 2013. M. Hajibaba, and S. Gorgin, “A Review on Modern Distributed Computing Paradigms: Cloud Computing, Jungle Computing and Fog Computing,” Journal of Computing and Information Technology, vol. 22, pp. 69-84, 2014. M. Abdelshkour, “IoT, from Cloud to Fog Computing”, March 2015, [Online] Avaialble: http://blogs.cisco.com/perspectives/iot-from-cloudto-fog-computing I. Stojmenovic, and S. Wen, “The Fog Computing Paradigm: Scenarios and Security Issues”, in Proc. of the 2014 Federated Conference on Computer Science and Information Systems, 2014. S. Depuru, L. Wang, and V. Devabhaktuni, “Smart Meters for Power Grid: Challenges, Issues, Advantages and Status,” Renewable and Sustainable Energy Reviews, vol. 15, pp. 2736-2742, Mar. 2011. M. A. Cusumano, “The changing labyrinth of software pricing”, Communications of the ACM, vol. 50, pp. 19-22, Jul. 2007. J. T. Ma, Y. M. Liu, and L. F. Wu, “New Energy Management System Architectural Design and Intranet/Internet Applications to Power Systems,” International Conference on Energy Management and Power Delivery, pp. 207-212, 1998.