Creating Energy IoT Solutions Using Intel SoC FPGA Devices and AWS Services
By Joseph Glover, Sr. Solutions Architect at AWS
By Arun Sehgal, APN Lead – Power and Utilities at AWS
By Kam Lee, IoT Technical Specialist at Intel
Power utilities around the world are headed towards the most radical shift in their business since their inception nearly 150 years ago.
The concepts of decarbonization and the democratization of assets are taking root in the energy supply business. At the same time, the digitization of the grid—along with the shifting requirements of regulatory bodies—creates new challenges and opportunities for utilities around the world.
Many utilities face an interesting challenge: improving customer engagement and customer retention. Historically, customer engagement with utilities has not been positive. It generally amounts to little more than the most basic of services, such as requests to connect or disconnect, send monthly bills, or provide outage reports.
In many parts of the world, utilities want to add new streams of revenue—beyond the mere sale of electrons. This goal necessitates a better understanding of end customers and their energy consumption patterns.
It’s crucial to provide a positive customer engagement mechanism that connects with the mindset of the current generation of energy consumers: “Alexa, pay my energy bill,” “Alexa, why is my energy bill so high?” or, more interestingly, “Alexa, charge my EV at 7:30 p.m. or whenever electricity is cheapest.”
Going a step further, utilities want to present customers with actionable information using Alexa skills. For example: “John, I notice that your air conditioner is performing below its standard efficiency. Do you want me to schedule maintenance with [utility name]?” These are the kinds of interactions that improve the stickiness of consumers to their utility companies.
Intel and Amazon Web Services (AWS) are committed to providing the tools, platforms, and services to make these capabilities a reality. In that context, this post introduces the concept of Energy IoT.
Energy IoT Opportunities
The ubiquity of internet connectivity and advances in cloud computing have propelled the concept of the Internet of Things (IoT) based on both data collection from edge and data analysis in the cloud. They have brought new applications and services to many verticals across consumer, commercial, and enterprise spaces.
In the energy vertical, it’s possible using energy disaggregation algorithms to determine the energy consumption behavior of individual appliances in a household by analyzing their main power signal.
This Energy IoT concept creates opportunities for solution providers along with utilities companies to deliver value-added services to consumers:
- Energy usage advisory: Solution providers can provide energy consumption statistics to consumers with a spending breakdown by appliance, alerts on inefficient usage or faulty appliances, and recommendations for specific ways of improving usage.
- Demand response/energy efficiency programs: Utility companies can forecast energy demands based on past and present energy consumption data. When demands are expected to go up, customers are notified to shut down non-essential appliances in the household. This conserves energy and avoids brownouts on the power grid. A rebate can be awarded to customers who participate in the program.
- In-place aging: Imagine an elderly parent living alone. Interested parties are alerted in a timely manner of abnormal energy usage or unusual appliance activity at the elderly person’s home. For example, a microwave oven might go unused for longer than the normal downtime pattern.
The technical solutions for these services can be readily implemented in a cost-effective and scalable manner. This solution uses edge devices based on Intel’s System on a Chip (SoC) field programmable gate array (FPGA), and software components on the AWS Cloud platform.
The architecture of Energy IoT solutions is largely based on the “edge-to-cloud” model. It consists of on-premises edge devices interworking with cloud-hosted software functions.
Figure 1 – Energy IoT End-to-End Architecture.
In this case, the edge devices are the “energy gateways” deployed in individual households. These are responsible for the following:
- Data acquisition: Monitoring main power lines to collect high-frequency samples of the power signal in real-time.
- Data transmission: Securely and efficiently sending measured power data to the cloud.
The energy gateway may also communicate with smart appliances (for example, W-Fi lamps or smart thermostats) in the household to control or regulate their operation based on commands received from the cloud.
In the cloud, data arriving from each household follows this task flow:
- Save to the database.
- Forward to a real-time visualization application.
- Feed into a processing pipeline for energy disaggregation and data analytics.
Optionally, the energy gateway may expose an API to provide third-party applications access to the data. For example, API access is needed when a utility company focuses only on data aggregation. They might rely on partners to perform energy disaggregation analytics and derive value from the data.
Energy Gateway Based on Intel CycloneV SoC FPGA
For mass deployment to households, the energy gateway needs to be low-cost and lightweight. Its power dissipation must also be low enough to meet the thermal constraints of the household power panel on which the gateway is mounted. These two design criteria can be met with an Intel CycloneV SoC FPGA, while fulfilling the performance requirements of the energy gateway.
An FPGA is an integrated circuit that can be programmatically hard-wired to implement digital logic and machine algorithms at high speed. Intel offers a family of SoC FPGAs that combines FPGA fabric with a host processor on a single chip. The CycloneV family delivers an excellent balance of performance, thermal mitigation, and cost for embedded hardware.
In particular, the CycloneV’s FPGA fabric is capable of high accuracy and a high sampling rate over two simultaneous analog-digital channels. It’s a good fit for acquiring and processing data from power lines.
Software can run on the CycloneV’s host processor to implement application logic and handle cloud communications through its built-in Ethernet interface or external Wi-Fi and Bluetooth interfaces. The diagram in Figure 2 shows the system on a chip with FPGA and built-in processor.
Figure 2 – Intel FPGA SoC consists of a programmable gate array and host processor.
Processing Energy Data Using AWS
The AWS Cloud delivers the scalability, security, and reliability required for Energy IoT solutions. Moreover, AWS provides building blocks that simplify the creation of data pipelines to process, store, and analyze IoT data.
These building blocks include:
- AWS IoT Core: An IoT connection service that can be used for secure transport of data and control commands between gateway and cloud, and for deployment and execution of AWS Lambda functions at the gateway.
- Amazon S3: A data “disk” service suitable for storing incoming data and hosting files.
- Amazon Kinesis Data Firehose: A data forwarding and buffering service that can be used to feed IoT data to other AWS components.
- Amazon Aurora: A relational database service that can be used for persistent storage and querying of IoT data.
- Amazon API Gateway: An API management service that can be used for implementing REST interfaces to allow retrieval of IoT data by third-party applications.
In addition, AWS supports integration with Alexa Voice Service (AVS). Energy IoT solutions can leverage this synergy to support smart-home use cases involving voice interactions. For example, a user may query Alexa to check if the stove is turned on, or request that Alexa turn on the air-conditioner using the energy gateway.
Proof of Concept
This section describes the prototype implementation of an end-to-end energy IoT solution based on the use of an Intel CycloneV-based energy gateway and AWS Cloud services.
On the local side, the energy gateway uses a current transformer to monitor a power line and can wirelessly control a Wi-Fi lamp. On the cloud side, it runs AWS IoT Greengrass software to securely connect to AWS IoT Core.
The energy gateway executes separate Lambda functions to support two use cases:
- Real-time visualization of power line data.
- Interworking with AVS for home power management.
In the first use case, data captured from the power line is periodically sent by the first Lambda function to AWS IoT Greengrass. Data arriving in AWS IoT Greengrass is cached in an Amazon S3 bucket. In addition, through Kinesis Data Firehose, the data is processed and saved in an Aurora database.
An API hosted on Amazon API Gateway enables a dashboard application hosted on Amazon S3 to retrieve the data for visualization. The dashboard shows current and historical energy consumption, enabling the energy customer to see a graphical representation of usage patterns. The following diagram shows this first use case.
Figure 3 – Use of AWS IoT Greengrass on FPGA SoC to publish energy data to AWS.
In the second use case, two Alexa Skills are set up: one to support querying the on/off state of the appliance connected to the power line being monitored, and another to support voice-based control of the Wi-Fi lamp.
In the case of the first skill, AVS returns an audio response based on cached data pulled from the Amazon S3 bucket. With the second skill, AVS invokes AWS IoT Core to issue an on/off command back to the energy gateway. The second Lambda function parses the command and triggers the appropriate action from the Wi-Fi lamp. The following diagram shows this second use case.
Figure 4 – Interworking of Alexa Voice Service and AWS IoT in Energy IoT solution.
By applying IoT concepts, solution providers, and utility companies can provide value-added services to help consumers conserve energy, improve the operation of household appliances, and realize demand response savings.
Such smart energy solutions can be implemented at scale by using low-cost energy gateways based on the Intel CycloneV SoC FPGA and by leveraging AWS services, including AWS IoT Core and AWS IoT Greengrass.
Intel – APN Partner Spotlight
Intel is an AWS IoT Competency Partner. AWS and Intel share a passion for delivering constant innovation. Together, we have developed a variety of resources and technologies for high-performance computing, big data, artificial intelligence/machine learning, and IoT.
*Already worked with Intel? Rate this Partner
*To review an APN Partner, you must be an AWS customer that has worked with them directly on a project.