An Expectation-Based Dynamically Adjusted Gossip Optimization Algorithm

The Gossip algorithm is a decentralized commu-nication algorithm primarily used to address information dis-semination and consistency issues in distributed systems. The Gossip protocol, with its significant distributed characteristics and advantages, has been widely applied in critical domains such as fault detection, resource allocation, global search, and robot operating system node communication. However, compared to traditional non-decentralized algorithms, the Gossip algorithm tends to generate more message redundancy during the infor-mation dissemination process, which to some extent impacts its communication efficiency. To address this issue, we propose a universal expectation-based dynamically adjusted Gossip op-timization algorithm that optimizes the message redundancy in the Gossip algorithm during the information dissemination process. This method effectively controls message redundancy by dynamically adjusting the number of message forwarding, denoted as K, in the Gossip algorithm. The experimental results show that in the three scenarios we compared, the message redundancy was reduced by 16.40% to 46.06%, and the original characteristics of the Gossip algorithm were not compromised. The code and experimental results in the paper are all open source.