In general, the higher the CPU frequency and the amount of RAM, the better. If the hardware is insufficiently powerful for the training process, all services on the host may stop responding.
For up to 600 users, a dedicated CPU with the following characteristics should fit the task:
- At least 2 GHz, quad-core.
- Support for Intel® MKL-DNN.
For training and positioning, you can use Wi-Fi hotspots / access points, Bluetooth beacons (with iBeacon support) or both.
On iOS devices, only iBeacons can be used.
Make sure the Wi-Fi/Bluetooth devices are whitelisted (MAC address or UUID for iBeacon).
- Make sure the GPS module is enabled on Android devices.
- Devices with Android 8 may have issues working with the feature in Wi-Fi mode.
- Devices with Android 9 don’t work with the feature properly.
- On Devices with Android 10, disable Wi-Fi scan throttling in the Developer options.
- Plans of levels within the location in JPEG format are required.
- If DeepMap is required, make sure you have the appropriate ZIP file.
- If calibration is needed, the GPS coordinates of the location are required. If DeepMap is available, the coordinates can be taken directly from the T.Flex for Android application.