How was the batch size chosen?

The batch size is chosen based on several factors:

• Device memory limitations. Embedded systems often have limited RAM, so the batch size must fit within the available buffer.
• Performance and response time. A larger batch can improve data processing efficiency but will increase latency.
• Characteristics of the data source. The frequency and volume of incoming data influence the optimal size.
• Power consumption. In IoT devices, minimizing energy consumption is important, so the batch size is selected considering transmission and processing optimization.

Often, the batch size is determined experimentally: starting with a small value, then increasing it to evaluate the balance between performance, latency, and resources. For example, if a device collects sensor data, it can accumulate 10-50 measurements before sending to reduce communication frequency and save energy.