Identifying Invalid Nodes in Bluetooth Mesh Networks by Using the Bayesian Probability Classification Algorithm and Time to Live

Abstract

Bluetooth mesh networks improve existing star networks by overcoming the problem of network service unavailability during single-node failure and enabling transmission for multiple Bluetooth devices. The flooding method does not consider transmission paths and merely transmits messages using advertisement channels (Channels 37, 38, and 39) to all nodes within its range using random paths in Bluetooth mesh networks. Time to live (TTL) and the Bayesian probability classification algorithm calculated the probabilities of the paths to identify invalid nodes. Invalid nodes influence overall network transmission performance. This study examined transmission time in single-path and multipath topologies for each radio channel (37, 38, and 39) and invalid nodes in a Bluetooth mesh network. Their roles were adjusted to improve network performance. The experimental results showed that replacing invalid nodes decreased the average return time by 18% for the single-path topology. Similarly, for the multipath topology, the modification of invalid nodes reduced the average return time by 27%.