Device Management for the Internet of Things
Writer: Dr. Ludo Cuypers, managing director COMmeto
Sensors play a crucial role in many use cases in the NextPerception project. As an integrator for IoT (Internet of Things) technology, COMmeto researches the integration challenges for data communication and device management. This research is applied in the “UC1-P1 Patient Monitoring” use case. This use case is carried out together with KULeuven, Televic and Pozyx. More details can be found in the public deliverable “D4.5 First Pilot Evaluation Reports”.
The figure below shows the general architecture of the network communication in the use case.
The sensor data generation starts from the left. Data are captured by sensor subsystems and preprocessed at the sensor level. The sensor data are then sent to the gateway device, where the data from different sources are synchronized and where further processing is applied. For example, in the UC1-P1 use case the processing of low-level movement data towards activity classification is likewise done. From the gateway device the data is sent to a broker subsystem from where the data is sent to the cloud infrastructure. From there the data can be used by other applications or consulted by humans.
On the top of the diagram there is the connection with the device management. The device management can be seen as an orthogonal dimension on the sensor data communication (referred to as application management).
To understand the need to differentiate device management from application management, assume the measurement of medical data. The medical doctor cares about the measured data and the processing of these data. However, the medical doctor does not want to (and should not) be involved in technical aspects like network connectivity, payment of subscriptions, device updates, battery levels, etc. Vice versa, technical personnel care about those aspects, but should not have access to the medical data as such. Therefore, device management and application management should be handled in a distinguishable way.
As the number of IoT devices increases, device management becomes more and more important while it also creates new challenges. Here are some of them.
In the Internet of Things, billions of devices need to be managed. Although processing happens in a distributed way, management must be possible in a centralized way (using distributed management components). The whole data collection and processing chain must be configurable for each single device or device class. New processing components, including evolvable artificial intelligence algorithms, need to be installed and updated on devices distributed in the network (for example on gateways). The support for the huge amount of individual end users (customers) must provide a good customer experience at a low operational cost. Further, all relevant kinds of sensor connectivity technologies must be supported, like wired protocols, Wi-Fi, 5G, narrow-band-IoT, Bluetooth, ZigBee, LoRaWAN, UWB etc. In addition, unpowered devices (battery, sun-powered, energy harvesting devices etc.), mobile devices (implying changing location and network connectivity) and devices without compute power (using proxies) must be taken into account.
Further, both device management and application management aspects should be handled in different life-cycle stages including device on-boarding, installation of new services, updates, new & changing configurations, maintenance, diagnosis and decommissioning (end-of-life management). All communications should apply state-of-the-art privacy and security recommendations.
To guarantee interoperability in a multi-vendor environment, the adherence to well defined standards is necessary. The standard that COMmeto examines in NextPerception is USP (TR-369) as defined by the Broadband Forum (https://usp.technology/). This standard builds further on the well accepted telecom standard TR-069.
The network nodes that need to be managed are called “agents”. The managing nodes are called “controllers”. Device management happens through agent-controller interaction. An agent contains a data model tree with objects and parameters. The objects in the data model represent proxied elements, network interfaces, software modules, managed services, IoT capabilities etc. By changing the data model object structure, or by setting object’s parameters, or by applying object’s commands, the controller determines the behavior of the agent. Further, the controller can define events on the object model and create subscriptions to them. This allows for agents to notify the controller on particular events happening.
The NextPerception project not only researches novel sensing and data processing techniques, but also explores how the data can be communicated in a scalable IoT network and how the IoT components can be managed in an increasingly growing IoT ecosystem. For more information on IoT integration and IoT device management, please contact COMmeto.