Edge computing and 5G networks are transforming the IoT, enhancing the capabilities of a wide range of applications across multiple sectors, and enabling a plethora of new and emerging ones. In this article, we illustrate the power of edge computing by examining three prominent use cases.
Industry 4.0 smart factory
Industry 4.0 offers manufacturers a vision of increased productivity and profitability based on flexible industrial environments that leverage smart technology and require limited human intervention. However, this transformation is hampered by the predominance in complex production domains of wired Ethernet networks, implemented because of their well-proven time-sensitive networking (TSN) capabilities.
Consider, for example, a modern assembly line that is no longer linear but adaptively routes work between cells based on capacity or the efficient warehouse that relies on automated guided vehicles (AGVs). These scenarios require the low latency, high-reliability wireless network connectivity offered by 5G networks.
By replacing existing hard-wired networks, 5G enables wide-ranging flexibility on the factory floor and brings deployment options, such as edge computing, software-defined networking (SDN), and private networking.
The combination of edge computing and private networking addresses the three overarching requirements of industrial automation – data locality, reliability, and low latency. With regulators around the world making shared spectrum available at a reasonable cost, factory operators no longer need to rely on the cloud and public, shared infrastructure to deliver their networking requirements. Sensitive production data can remain local to the factory or plant. Open RAN (O-RAN) and SDN networking functionality enable the cost-effective installation of powerful, local 5G radio access networks (RAN). These small-scale RAN networks can be flexibly deployed and configured to meet the latency, reliability, and availability requirements of the specific industrial process. Existing wired networks that control lower latency applications, such as motors, drivers, and actuators, can be integrated with the 5G RAN infrastructure, mapping relevant TSN features to 5G, figure 1.
Connected automobile In many ways, the connected vehicle can be considered as the ultimate IoT edge device, combining multiple, high-performance edge compute nodes with high-bandwidth networking and dozens of sensors. Sophisticated hardware and software architectures are emerging to replace the multiple electronic control units (ECUs) found in high-end automobiles. These new architectures are based on a centralized compute resource, the service-oriented gateway (SoG) based on high power, multicore SoC products, such as the NXP S32G vehicle network processor. The SoG communicates with the domain controllers, which are responsible for the main functionalities within the vehicle, figure 2, and provides a secure path to the cloud via the onboard telematics control unit (TCU).
The vehicle edge processing enabled by the power of the SoG is fundamental to the evolution of the connected vehicle. The terabytes of raw data generated each hour by the vehicle’s sensors can be processed locally, and decisions can be taken in real-time based on information and anomalies detected by the sensors. Sophisticated machine learning (ML) algorithms can be run at the vehicle edge, enabling enhanced levels of intelligence and autonomy by predicting and reacting to a wide range of driving conditions.
The SoG also orchestrates vehicle-wide over-the-air (OTA) updates of software, firmware, and ML models, which continually improve the vehicle’s in-life performance, while supporting incremental revenue-generating services for the vehicle OEMs.
The roll-out of 5G is seen as essential to the further development of the autonomous vehicle as its ultra-low latencies, combined with multi-access edge computing (MEC), move cloud servers away from the data center, closer to the vehicles at the network edge. The resulting reduction in latencies to single-digit milliseconds enables data center levels of processing power with the real-time responsiveness required by the autonomous vehicle.
The ever-increasing power and shrinking form factors of electronic devices enable a rapid expansion in the variety and sophistication of use cases for wearables. Wearables can come in the form of implantable devices, fitness trackers, smart jewelry, watches, shoes, and clothing, with emerging applications such as:
- Health monitoring, with movement and fall detection, emergency calling, and vitals reporting (for example, body temperature, oxygen levels, and cardiac signs)
- Sleep monitoring and therapy
- Meal/calorie tracking
- Hearing enhancement
- Social distancing, and contact tracing and tracking
The compute power contained in wearable devices, along with the capabilities of peripheral devices and sensors, enable them to function as true edge processing devices. These technological advances are driving the convergence of wearables with smart home applications, delivering increased levels of convenience to consumers. Smart watches can now be used to control everything from lighting systems to coffee machines and can also control entry and security systems. Smart watch applications are also being integrated with connected car tools, enabling user access to controls such as sunroofs, seat ventilation, temperature, and status alerts.
Battery life is critical in wearables; low-power operation is essential with battery sizes restricted by physical space limitations. The modern edge processors within wearables can support a variety of reduced power modes, using various low-power design techniques to achieve extended battery lives in both active and sleep modes.
Alongside the increased levels of processing power, developments in sensors and peripheral devices enable enhanced functionality in wearables, including sophisticated graphics, localization, voice control, and motion sensing. The edge computing capabilities of wearables also ensure the rapid and secure processing of the vast quantities of data generated, reducing the need for broadband processing, lowering power consumption and, hence, increasing battery life.
Edge processing is removing the limitations of cloud computing
With the rapid expansion of the IoT, the increased numbers of connected devices have exposed the limitations, particularly the high latencies of the classic cloud computing model. Although edge computing is not a new concept, its application to embedded edge devices has been made possible by the increased processing power and shrinking form factors of the electronics contained in these edge devices. As the power and sophistication of these edge devices continue to increase, existing applications will be enhanced, and new ones will emerge.
Robert Oshana is vice president of software engineering R&D for the edge processing business line at NXP Semiconductors. He serves on multiple industry advisory boards and is a recognized international speaker. He is chief editor of Essentials of Edge Computing (2022), and has published numerous other books and articles on software engineering and embedded systems. He is also an adjunct professor at Southern Methodist University and is a Senior Member of IEEE.
- Edge computing platform enables deep access to automotive data
- Defining a standard federated model for multi-access edge computing
- Edge computing is here: what’s next?
- How edge supercomputing will transform the great data deluge
- Arduino adds standalone intelligent camera for edge machine vision
- Software-defined processors for edge AI get two new compilers