Contextual Gossip Protocol for Dynamically Changing Networks

Gossip protocols are utilized to communicate information over a wide network of nodes. In a gossip protocol, a network node periodically exchanges information with a subset of nodes such that information hops across nodes to reach the ends of the network. In a dynamic network, a message-passing protocol that is not cognizant of the current network topology results in inefficient message propagation. This disclosure describes an efficient gossip protocol in which a global network is dynamically divided into subnetworks. Each message-transmitting node detects the subnetwork it currently belongs to. The node sets a higher priority of message-passing to nodes within its subnetwork. Subnetwork determination can be done by feeding user-permitted contextual information including motion, location, etc. to a trained machine learning model. The described techniques result in efficient message propagation