Performance evaluation of device discovery in bluetooth LE with probabilistic model checking

The device discovery in Bluetooth is needed before establishing data communications. Almost all Bluetooth slaves are battery-powered so that they should connect with a master through the device discovery for a short time. To prevent the worst case scenario with long device discovery in Bluetooth, the events with long device discovery should be estimated in advance. However, it is intractable to estimate the exact longest device discovery time with Monte Carlo simulations or mathematical analysis because of the probabilistic behavior. In this paper, we reveal the longest device discovery time and the maximum energy consumption in Bluetooth Low Energy (LE) with a probabilistic model checking tool, PRISM. The results show that some parameters in Bluetooth LE will largely affect the energy consumption so that their design criteria will be desired to reduce the energy consumption.