Note:
If a version older than v1.0.0 is already installed on the Industrial Edge Device, it must be uninstalled before installing the AI Asset Manager.
Uninstalling the app will result in the loss of all data in the AI Asset Manager.
The following image shows the AI Asset Manager and the software systems around it:
To which it establishes connections
Which can access its API
HW configuration
Recommended HW configuration:
Other supported HW configurations:
Industrial Edge - SIMATIC IPC227E
Industrial Edge - SIMATIC IPC427E
Industrial Edge - SIMATIC IPC847E
Compatible applications
AI Inference Server: version ≥ 1.3.0 (partially compatible, Connection Upgrade not supported)
AI Inference Server variants: version ≥ 2.0.0 (Connection Upgrade supported):
AI Inference Server
AI Inference Server - 3 pipelines
AI Inference Server GPU accelerated
Monitoring functionalities only work with AI Inference Server variants ≥ 2.4.0
Industrial Edge platform
IED OS: version ≥ 1.8.x
Edge device settings
Network Time Protocol (NTP) synchronization
For certain services of the application to work properly, the Edge device on which AI Asset Manager is installed and must have the correct time settings. The status of NTP synchronization can be checked in the Edge device's UI in the Statistics menu. If the NTP settings need to be modified, please refer to the relevant sections of the Industrial Edge Management Operation and Industrial Edge Device Operation manuals.
Edge Management timer
AI Asset Manager sends deployment-related actions through the central IE Management system, so for lower latency in the communication chain, it is recommended to change the default setting of Edge management timers on Edge devices that have an Edge application with AI Asset Manager Agent installed (i.e. AI Inference Server). The time for retrieving jobs from the IEM can be set in the "Check Edge Management Notifications" setting. The recommended value for this is 10 seconds.
Network and firewall configuration
The Industrial Edge Device must be set up according to the required network settings.
The required setting can be found at Industrial Edge Security overview, chapter 2.3.2 "IP Protocols and Ports".
As shown in the figure, the AI Asset Manager API must be accessible from all Industrial Edge devices (IED) that are the target of the deployments (target devices). The IED (physical or virtual) on which AI Asset Manager is installed (Manager device) may have several network interfaces. One of them must be accessible from the target devices using the assigned IPv4 address. If the manager device and the target devices are connected to different networks, there must be a network route to enable connection.
If you are installing AI Asset Manager on a virtual device, we recommend to configure a Bridged Network Adapter for the virtual machine. Then the virtual network adapter can have an IP address assigned from the host network.
In addition to the required network settings for Industrial Edge, the following ingress and egress rules must be implemented in the firewalls for the listed features to work.
| Port | Protocol | Direction | Usage / Feature |
|---|---|---|---|
| 443, TCP | HTTPS | Ingress | Application UI, API Integration workspace |
| 8883, TCP | MQTTS | Egress | Workspace with AWS IoT Core or Azure IoT Hub integration if the connection is established using MQTT protocol over TLS transport |
| 443, TCP | MQTTS | Egress | Workspace with AWS IoT Core integration if the connection is established using MQTT protocol over TLS and Application Layer Protocol Negotiation (ALPN). This rule applies to Siemens AI Services integration as well |
| 443, TCP | WSS | Egress | Workspace with Azure IoT Hub integration if the connection is established using MQTT over Secure WebSockets |
| 443, TCP | HTTPS | Egress | Downloading Packages from Cloud storage services |
The IED where the AI Asset Manager is installed needs to reach the Azure IoT Hub.