At Interop 1990, Internet pioneers John Romkey and Simon Hackett connected a Sunbeam Deluxe toaster to the global TCP/IP network using a Simple Networking Management Protocol Management Information Base (SNMP MIB) to control the power switch. In doing so, they produced the world’s first piece of ‘Internet bread.’ At Interop 1991, Romkey and Hackett added a robot that picked up a slice of bread and dropped it into the toaster, automating the system from end-to-end. This was the world’s first Internet of Things (IoT) application.
The ‘Internet toaster’ is an example of a basic three-layer IoT architecture: device, connectivity, and application. The device layer, at the bottom, is the source of IoT data. The middle connectivity layer exists to provide services such as network access and data transmission to the device. And, the top applications layer implements processing, storage, analysis, and presentation in addition to service support and operations management.
Through the early 2000s, the IoT advanced on connectivity issues with no significant changes at the application layer. Disparate, proprietary solutions led to expensive systems and the types of fragmentation that can occur when technologies are trapped in restricted vertical markets.
The emergence of cloud computing has especially impacted the application layer. A wide range of cloud products integrated with the latest ICT technologies are now providing a diverse range of services. A key driver for this new generation of IoT products is the significantly lower cost for application development, deployment, and operational support. Additional benefits include shorter service provisioning times and less need for highly skilled technicians. IoT applications have become both simpler and more efficient.
Since 2012, the United States, Germany, and China have proposed their respective long-term plans to introduce IT technology to the industrial sector, namely, the Industrial Internet, Industry 4.0, and Made in China 2025. These strategies use IoT for connecting hundreds of billions of devices and sensors to networks. The resulting connectivity will allow the exchange of data and information between machines, and between machines and people.
The IoT development architects at Huawei have identified a number of challenges with the basic cloud-device system topology:
Timeliness: Round-trip latencies from the edge sensor to the cloud processor and back often exceed the millisecond and microsecond response times required for process control, analysis, and/or decision-making functions at many industrial and commercial sites.
Security: Unprotected data between edge devices and cloud processing centers leave trade secrets vulnerable to hackers.
Reliability: Production processes and control instructions depend on highly reliable data that risk being disrupted over long service chains.
Congestion: Large numbers of IoT nodes may generate enormous amounts of data that can introduce congestion over network data links and in processing queues. The delay, jitter, and data loss caused by the congestion create negative impacts on industrial efficiency.
Cost: Higher operating expenses are incurred if all processing and storage of edge-device inputs is transferred to cloud data centers.
IoT System Architecture
Not a new concept, edge computing deploys time-, security-, and reliability-sensitive applications onto network edge nodes (such as gateways) that are physically sited near sensors and data sources. This deployment strategy allows for captured or generated data to be processed and analyzed in the shortest time and over the smallest distance. Edge computing platforms enable local data aggregation and storage to reduce network congestion and lower cost.
Edge computing has the following five key benefits:
Large numbers of sensor and terminal connections at network edges
Support for time-sensitive services with microsecond responses from network edge gateways
Data aggregation eliminates fragmented data, shields noise, and allows processed data to be uploaded on demand
Intelligent analysis at the network edge gateways enables flexible service adjustment and automated Operations and Maintenance (O&M) processing
Deep protection of private security zones provides an end-to-end, field-specific, and customizable protection mechanism for equipment, networks, and data
Edge computing supplements cloud computing. Application functions need to be decoupled in a service-oriented manner based on specific service scenarios, and be deployed in a dynamic and distributed fashion between cloud and edge nodes to achieve an optimal balance between costs and user experience.
The Huawei Edge Computing-IoT (EC-IoT) solution uses Software-Defined Networking (SDN) to manage large numbers of IoT network devices to achieve the following:
Isolation of edge gateway computing, storage, and I/O resources
Dynamic loading of edge services
API exposure to basic device capabilities
Heterogeneous access and real-time interactions with IoT terminals
Built on Huawei’s Agile Controller and open edge gateways, the EC-IoT solution enables large numbers of IoT gateways to trade comparatively low computing performance for the benefit of the local processing, storage, and communications needed to mediate between terminal connections and cloud data centers. The EC-IoT solution helps shift networks from cost centers to business value centers with the following characteristics:
Each industry has unique requirements for its IoT applications. For example, applications for simple meter collection and analysis are best deployed from a cloud data center. However, in the domain of industrial process control, applications must operate in on-site gateways to resolve the necessary levels of real-time analysis and control policy association for the collected data. In complex manufacturing scenarios, different application services should be deployed onto corresponding local gateways, enterprise data centers, cloud platforms, and smart terminals, and configured to work collaboratively.
A key feature of the EC-IoT environment is the ability to deploy different types of applications based on specific service requirements. Further, the IoT platform provides full lifecycle management and real-time monitoring of network-wide activation status, and health, data collection, and convergence ratios. Collectively, these functions play a pivotal role in optimizing end-user services.
EC-IoT uses various Virtual Machines (VMs), containers, and other machine constructions based on x86/ARM processors running the Linux operating system. This platform allows users to run open-source applications that decouple the development of their operating environments at minimal cost.
EC-IoT abstracts the capabilities of edge devices via Application Programming Interfaces (APIs) for network interfaces, data collection and storage, message subscriptions and releases, and local policy engines. Third-party middleware can be integrated with network devices to provide IoT application services through a shared bus. The Huawei EC-IoT solution leverages these features onto a general platform that enables the rapid development of customer-specific applications.
Compared with the Internet, the types of IoT terminals is much more diverse. Adding up to tens and hundreds of billions of sensors, terminals, communication modules, SIM cards, IoT gateways, and management nodes, the network management and security tools used for the Internet do not fulfill requirements for such huge numbers of IoT connections and nodes.
The core competitiveness of the IoT industry relies on fast service innovation and comprehensive user experiences. Providers that invest in the necessary levels of service development and operational support expect that the rewards will far outweigh the costs.
Huawei EC-IoT framework
Based on the architecture, EC-IoT data plane connections are selectable for specific services. The same service may assigned to use different data connections based on the Service Level Agreements (SLAs) of different customers. Further, IoT data connections can be upgraded or reconstructed as services are extended into other markets or facilities.
With the growing sophistication of the cloud computing industry, IoT platform services can be selected based on the ability of the providers to offer more choices — and lowering the barriers to entry for new IoT customers based on the growing use of cloud migration tools.
EC-IoT gateways isolate the management of end-point network connections from the cloud service applications, which allow users to select cloud services by themselves without having to know or understand the technical details of connectivity. Because the EC-IoT horizontally decouples the vertical IoT architecture, the IoT network infrastructure is unbound from the cloud platform. The result opens the entire industry for the greatest number of users by reducing upfront investments and O&M costs.
The Edge Computing Consortium (ECC) was established in Beijing, China in November 2016 as a joint initiative by Huawei Technologies, Shenyang Institute of Automation of Chinese Academy of Sciences, China Academy of Information and Communications Technology (CAICT), Intel Corporation, ARM, and iSoftStone. As of 2017, ECC has over 100 members.
EC-IoT has entered service for the electric power industry to support the long-term evolution of electricity distribution. In the retail sector, EC-IoT platforms are innovating the operating models of chain stores and vending machines. For urban development, EC-IoT is accelerating the platforms for security surveillance, energy conservation, and smart transportation. In agriculture and stockbreeding, EC-IoT is enabling IT-powered production initiatives. For industry, EC-IoT is converging ICT and Operational Technology (OT) to achieve predictive device maintenance. For warehouse logistics, smart homes, and cold chains for food security, EC-IoT is a popular and innovative technical investment. An increasing number of EC-IoT applications are being proven in testbeds and commercial use. In summary, EC-IoT is a powerful engine for realizing the goal for the IoT to enable the connectivity for hundreds of billions of networked devices.