Wildcard Topic Management Using Bloom Filter in Distributed MQTT Brokers

MQTT is a widely used protocol in IoT systems. Due to the central role of an MQTT broker in handling messages, large-scale systems need to distribute the load across multiple brokers. When using multiple brokers, sharing subscription topics among them is a common strategy to prevent message flooding. Given the potentially large volume of topics, some existing approaches use a Bloom filter to store and manage them with space-saving. However, these methods have difficulties in managing wildcards, i.e., many queries to the Bloom filter could occur to search one topic. In this paper, we introduce a method to handle wildcard topics efficiently using a Bloom filter. The proposed method adds prefixes of subscription topics to the Bloom filter. Searching for a published topic is processed like using Trie structure while considering wildcard patterns. Experimental results demonstrate that our method reduces the number of queries to the Bloom filter compared to existing methods in many cases.