This thesis focuses on the design and implementation of the wireless localization module, using UWB technology with emphases on low-power firmwre based on RTOS. Wireless localization is based on TDoA algorithm.  The resulting HW module is designed as a four layer PCB, based on MCU crf52832 (ARM Cortex M4) and UWB module DevaWave DW1000. Firmware is implemented using FreeRTOS with emphasis on low power consumption. For hardware implementation, Eagle CAD was used. Firmware is implemented in C and Assembly programming languages.

When choosing what IoT protocol to use today, there are lots of choices. If a mesh type network is chosen, Bluetooth mesh might be a possible candidate. Bluetooth mesh without correctly configured parameters can however suffer from congestion and packet loss if the network is very dense or consists of many nodes. This can be counteracted by choosing which nodes should be relays more carefully, as well as setting the re-transmission count and Time To Live (TTL) based on the current topology of the network. If the nodes in the network change position or are added/removed regularly, it is impossible to set the parameters optimal for all the possible network layouts. This is where a dynamic configuration comes in handy. In this master thesis a custom control model was created which implemented the K2 Pruning algorithm for relay selection, custom heartbeats for a dynamic TTL on each node, and a static re-transmission count for message originators and relays. A possible way to implement a dynamic re-transmission count is also discussed, as well as how the dynamic configuration could be autonomous without the need of physical interaction when reconfiguring the network. The implemented dynamic configuration tested on a physical system of 33 nodes was partly unsuccessful, but still provided improved Packet Delivery Ratio (PDR), reduced message delay, and useful knowledge for future implementations of a dynamic configuration. The K2 Pruning algorithm failed in choosing relays correct and quickly due to congestion during the neighbor information exchange needed to run the algorithm. Therefore, a different relay selection algorithm is suggested for future models or the refrain of acknowledged messages during the neighbor information exchange phase.