Parallel database algorithms for solving planning problems on very large databases

Abstract

SUMMARYThis paper discusses cooperative distributed planning problems using multiple agents that share a very large database. It is an important framework for planning by multiple agents with incomplete information, and for evaluating behavior strategies and negotiation manners. The specific issue involves distributed constraint-satisfymgproblems on databases. The problem is described as concurrent execution and repetitive execution of transactions, where each agent is deemed as a transaction that operates on a very large database. Therefore, we propose a highly efficient parallel database processing algorithm for these two basic operations.