BeeColony & BeeNation: Decentralized Swarm Intelligence in Integer-Only Neuromorphic Networks

Distributed artificial intelligence typically requires a central coordinator or parameter server, incurring heavy networking overhead and single points of failure. In this paper, we present BeeColony and BeeNation: two hierarchical layers that extend the MultiBeeSketch model to decentralized, peer-to-peer swarms. Operating completely with integer arithmetic (zero-floats), a BeeColony orchestrates localized worker swarms that exchange knowledge matrices through direct, sparse state-sharing. BeeNation acts as a macro-layer, routing symbolic inference and coordinate grids across a mesh of independent colonies.