The Internet of Things (IoT) is the latest buzzword driving the industry for any number of low-power interconnected things. However, IoT encompasses an incredible number of different types of things ranging from edge objects, namely smart or wearable devices which are battery powered with sensors and wireless connectivity, through aggregation nodes, namely hubs, routers and gateways for data aggregation, up to information processing servers in the Cloud to handle the data pushed by edge objects.
Whereas data aggregation and information processing in the Cloud requires high speed and/or high performance computing, a whole range of processing capabilities is required to cover the needs of diverse edge objects. Optimizing for low-power consumption and for Return on Investment according to edge object development cost vs. selling price requires making the best compromises on each subsystem architecture. This implies making the appropriate choice of microcontroller (MCU) or microprocessor (CPU), of their subsystem modes and power management network, per application requirements.
Several segmentations of edge objects can be made depending on the angle of analysis. A first approach is to segment their field according to operating systems, thus to the processing power requirements, where three different levels can be distinguished:
- Entry level needs with 8-16 bit MCUs running a dedicated application with no operating system,
- Mid range needs with 16-32 bit MCUs running a Real-Time Operating System (RTOS) of even a Java OS,
- High end needs with 32-64 bit Application Processors (AP) running an operating system such as Linux or Android to provide graphical data displays and intuitive user interactions.