Implementation Viewpoint
The Implementation Viewpoint must provide sufficient detail to have meaningful technical discussions between the different technical stakeholders of the product team. However, most design artifacts in this viewpoint are still on a level of abstraction which will hide many of the different details required by the implementation teams.
End-to-End Architecture
Asset Integration
The Asset Integration perspective should provide an overview of the physical parts of the product, including sensors, antennas, battery / power supply, HMI, and on-board computer. The focus is how these different elements are integrated with the asset itself. For example, where exactly on the asset would the antenna be located, where to position the main board, battery, etc.
Hardware Design
Depending on the requirements of the AIoT system, custom hardware development can be an important success factor. The complexity of custom hardware design and development should not be underestimated. Also, for smart, connected products, an issue with custom hardware is the supply side - either manufacturing the hardware internally, or finding a suitable manufacturer. For smart, connected solutions w, mass manufacturing is typically not as much an issue, because of the smaller volumes typically required. In this case, the required hardware can be produced in a smaller lab.
From the hardware design point of view, a key artefact is usually the schematic PCB design (Printed Circuit Board). See example for the smart, connected tightening system below. In order to get production ready, this will have to be detailed and eventually encoded in so called Gerber-files.
Software Architecture
The Software Architecture should provide a high-level view of how key software elements are supporting the product design. There are some re-usable AIoT Software Architecture patterns like Lambda and Kappa architectures. An example is shown below.
AI Pipeline Architecture
The AI Pipeline Architecture should explain on a technical level how data preparation, model training and deployment of AI models are supported.
