Distributed Dynamical Omnicast Routing

摘要

The omnicast problem addresses the need of uniform information dispersal in a group of individual nodes (robots), that is, every robot needs to receive (and process) the observations processed by everybody else. The time (or the number of schedule slots) it takes to exchange all distributedly collected information between all nodes/robots is a critical limiting factor for almost all practical swarm or distributed sensing applications. The establishment of a practical distributed scheduling scheme for this purpose is therefore crucial. Actual practical constraints of limited communication ranges, low bandwidth, asynchronous entities, and disturbances on the communication channels complicate this problem. This paper proposes a method of distributed dynamical omnicast routing (DDOR) which converges and solves the given problem, and does so with high performance ratings, as demonstrated in simulations. The problem of fast information dispersal in short range, and limited bandwidth communication systems in heterogeneous swarms of autonomous vehicles occurs in many applications. One example is the distributed control of large schools of underwater vehicles.