A hybrid P2P and pub/sub messaging system for decentralized Information Management

A key property in tactical network environments is that the wireless network itself is typically the major limiting factor in how much information can be shared between users. In such environments, tactical operations often employ group communication protocols (one-to-many or many-to-many) to send information between users. This often results in situations where every message sent gets delivered to everyone on the network which can lead to network congestion, performance degradation, application failure, and even complete information loss. In light of today's computationally rich mobile devices, it is preferential to expend additional processing power to limit the messages being sent to only those required by interested clients rather than to simply deliver every message to every client. This paper summarizes the results of a research effort to explore and implement an Information Management (IM) prototype known as BANDIT that enables efficient and reliable information exchange across a swarm of Android devices within a decentralized Peer-to-Peer (P2P) network. BANDIT features a lightweight implementation of the Kademlia algorithm for P2P swarm management, and implements an efficient broadcast algorithm designed for minimal cost internode communication within structured P2P networks. BANDIT provides decentralized publish/subscribe IM services that, in contrast to more traditional IM systems, have no reliance on a centralized server (which requires additional dedicated hardware and poses the risk of being a single point of failure). Instead, BANDIT IM services are provided by each of the devices participating in the swarm. BANDIT is interoperable with traditional client-server based IM systems which further enables information sharing across tactical and strategic elements.