The Internet of Things on AWS – Official Blog
Scaling for Complexity – Architecting for Performant Embedded Devices at the Edge – Part 2
The following is a survey paper, published and presented to the Academic Congress of Embedded World 2022 at Nuremberg, Germany on June 21st 2022.
Part 1 – Scaling for Complexity – Architecting for Performant Embedded Devices at the Edge – Part 1
The provisioning layer of your IoT workloads consists of the Public Key Infrastructure (PKI) used to create unique device identities and the application workflow that provides configuration data to the device. The provisioning layer is also involved with ongoing maintenance and eventual decommissioning of devices over time. IoT applications need a robust and automated provisioning layer so that devices can be added and managed by your IoT application in a frictionless way. When you provision IoT devices, you must install a unique cryptographic credential onto them. Typically, AWS IoT devices use X.509 certificates, while mobile applications use Amazon Cognito identities or other custom mechanisms.
By using X.509 certificates, you can implement a provisioning layer that securely creates a trusted identity for your device that can be used to authenticate and authorize against your communication layer. X.509 certificates are issued by a trusted entity called a certificate authority (CA). While X.509 certificates do consume resources on constrained devices due to memory and processing requirements, they are an ideal identity mechanism due to their operational scalability and widespread support by standard network protocols.
AWS Certificate Manager Private CA helps you automate the process of managing the lifecycle of private certificates for IoT devices using APIs. Private certificates, such as X.509 certificates, provide a secure way to give a device a long-term identity that can be created during provisioning and used to identify and authorize device permissions against your IoT application.
AWS IoT Just-In-Time Registration (JITR) enables you to programmatically register devices to be used with managed IoT platforms such as AWS IoT Core. With Just-In-Time Registration, when devices are first connected to your AWS IoT Core endpoint, you can automatically trigger a workflow that can determine the validity of the certificate identity and determine what permissions it should be granted.
The Communication layer handles the connectivity, message routing among remote devices, and routing between devices and the cloud.
The Communication layer lets you establish how IoT messages are sent and received by devices, and how devices represent and store their physical state in the cloud. AWS IoT Core helps you build IoT applications by providing a managed message broker that supports the use of the MQTT protocol to publish and subscribe IoT messages between devices.
The AWS IoT Device Registry helps you manage and operate your things. A thing is a representation of a specific device or logical entity in the cloud. Things can also have custom defined static attributes that help you identify, categorize, and search for your assets once deployed.
With the AWS IoT Device Shadow Service, you can create a data store that contains the current, last reported, and desired state of a particular device. The Device Shadow Service maintains a virtual representation of each of your devices you connect to AWS IoT as a distinct device shadow. Each device’s shadow is uniquely identified by the name of the corresponding thing.
With Amazon API Gateway, your IoT applications can make HTTP requests to control your IoT devices. IoT applications require API interfaces for internal systems, such as dashboards for remote technicians, and external systems, such as a home consumer mobile application. With Amazon API Gateway, you can create common API interfaces without provisioning and managing the underlying infrastructure.
A key business driver for IoT is the ability to aggregate all the disparate data streams created by your devices and transmit the data to your IoT application in a secure and reliable manner. The ingestion layer plays a key role in collecting device data while decoupling the flow of data with the communication between devices.
With AWS IoT rules engine, you can build IoT applications such that your devices can interact with AWS services. AWS IoT rules are evaluated and actions are performed based on the MQTT topic stream a message is received on.
Amazon Kinesis is a managed service for streaming data, enabling you to get timely insights and react quickly to new information from IoT devices. Amazon Kinesis integrates directly with the AWS IoT rules engine, creating a seamless way of bridging from a lightweight device protocol of a device using MQTT with your internal IoT applications that use other protocols.
Similar to Kinesis, Amazon Simple Queue Service (Amazon SQS) should be used in your IoT application to decouple the communication layer from your application layer for critical event processing. Amazon SQS enables an event-driven, scalable ingestion queue when your application needs to process IoT applications where message order is not required.
One of the benefits of implementing IoT solutions is the ability to gain deep insights and data about what’s happening in the local/edge environment. A primary way of realizing contextual insights is by implementing solutions that can process and perform analytics on IoT data.
IoT workloads are often designed to generate large quantities of data. Ensure that this discrete data is transmitted, processed, and consumed securely, while being stored durably.
Amazon S3 is object-based storage engineered to store and retrieve any amount of data from anywhere on the internet. With Amazon S3, you can build IoT applications that store large amounts of data for a variety of purposes: regulatory, business evolution, metrics, longitudinal studies, analytics machine learning, and organizational enablement. Amazon S3 gives you a broad range of flexibility in the way you manage data for not just for cost optimization and latency, but also for access control and compliance.
Analytics and Machine Learning Services
After your IoT data reaches a central storage location, you can begin to unlock the full value of IoT by implementing analytics and machine learning on device behavior. With analytics systems, you can begin to operationalize improvements in your device firmware, as well as your edge and cloud logic, by making data-driven decisions based on your analysis. With analytics and machine learning, IoT systems can implement proactive strategies like predictive maintenance or anomaly detection to improve the efficiencies of the system.
AWS IoT Analytics makes it easy to run sophisticated analytics on volumes on IoT data. AWS IoT Analytics manages the underlying IoT data store, while you build different materialized views of your data using your own analytical queries or Jupyter notebooks.
Amazon Athena is an interactive query service that makes it easy to analyze data in Amazon S3 using standard SQL. Athena is serverless, so there is no infrastructure to manage, and customers pay only for the queries that they run.
Amazon SageMaker is a fully managed platform that enables you to quickly build, train, and deploy machine learning models in the cloud and the edge layer. With Amazon SageMaker, IoT architectures can develop a model of historical device telemetry in order to infer future behavior.
AWS IoT provides several ways to ease the way cloud native applications consume data generated by IoT devices. These connected capabilities include features from serverless computing, relational databases to create materialized views of your IoT data, and management applications to operate, inspect, secure, and manage your IoT operations.
The purpose of management applications is to create scalable ways to operate your devices once they are deployed in the field. Common operational tasks such as inspecting the connectivity state of a device, ensuring device credentials are configured correctly, and querying devices based on their current state must be in place before launch so that your system has the required visibility to troubleshoot applications.
AWS IoT Device Defender is a fully managed service that audits your device fleets, detects abnormal device behavior, alerts you to security issues, and helps you investigate and mitigate commonly encountered IoT security issues.
AWS IoT Device Management eases the organizing, monitoring, and managing of IoT devices at scale. At scale, customers are managing fleets of devices across multiple physical locations. AWS IoT Device Management enables you to group devices for easier management. You can also enable real-time search indexing against the current state of your devices through Device Management Fleet Indexing. Both Device Groups and Fleet Indexing can be used with Over the Air Updates (OTA) when determining which target devices must be updated.
In addition to managed applications, other internal and external systems need different segments of your IoT data for building different applications. To support end-consumer views, business operational dashboards, and the other net-new applications you build over time, you will need several other technologies that can receive the required information from your connectivity and ingestion layer and format them to be used by other systems.
Database Services – NoSQL and SQL
While a data lake can function as a landing zone for all of your unformatted IoT generated data, to support all the formatted views on top of your IoT data, you need to complement your data lake with structured and semi structured data stores. For these purposes, you should leverage both NoSQL and SQL databases. These types of databases enable you to create different views of your IoT data for distinct end users of your application.
Amazon DynamoDB is a fast and flexible NoSQL database service for IoT data. With IoT applications, customers often require flexible data models with reliable performance and automatic scaling of throughput capacity.
With Amazon Aurora your IoT architecture can store structured data in a performant and cost-effective open-source database. When your data needs to be accessible to other IoT applications for predefined SQL queries, relational databases provide you another mechanism for decoupling the device stream of the ingestion layer from your eventual business applications, which need to act on discrete segments of your data.
Frequently, IoT workloads require application code to be executed when the data is generated, ingested, or consumed/realized. Regardless of when compute code needs to be executed, serverless compute is a highly cost-effective choice. Serverless compute can be leveraged from the edge to the core and from core to applications and analytics.
AWS Lambda allows you to run code without provisioning or managing servers. Due to the scale of ingestion for IoT workloads, AWS Lambda is an ideal fit for running stateless, event-driven IoT applications on a managed platform.
The AWS Well-Architected Framework provides architectural best practices across the pillars for designing and operating reliable, secure, efficient, and cost-effective systems in the cloud for IoT applications. The framework provides a set of questions that you can use to review an existing or proposed IoT architecture, and also a set of AWS best practices for each pillar.
Utilizing the concept of an IoT OSI Model stack architecture, a business workload can be decomposed into the seven topics of the AWS Well-Architected Framework for achieving business value at each of the seven OSI layers. This in turn can be aggregated as needed into the four constructs of business outcomes. Therefore, working backwards from the business outcomes warranted, the framework in the IoT architecture design helps produce performant, stable, and efficient systems ready for scaling, which allows a business to focus on its functional business requirements, which is key for growth.
About the author
Channa is a Specialist Solutions Architect, working on IoT and Robotics at Amazon Web Services (AWS) and part of the internal Technical Field Community for Telecom and IoT. Prior to AWS Channa has had an extensive career in Telecom, working with Tier 1 carriers around the world implementing various voice and messaging products for their SS7 networks. At AWS Channa works with Enterprise customers, and has built and presented IoT projects for re:Invent (2019/2020/2021), re:Inforce(2021/2022), Embedded World (2021/2022), and Hannover Messe (2021).