Network Flow Optimization Theory and its Distributed Algorithm based on Tie-set Flow Vector Space

摘要

As modern information network is rapidly becoming larger and more complex system, the most important issue is how to use network resources with the highest efficiency and to provide the maximum service to the customer. In order to solve the issue, the establishment of the method of distributed management system and efficient traffic control is essential. As one of the basic theory for establishment of the such the system, this paper dealt with network flow optimization theory based on their theory and presented distributed algorithms for various network flow optimization problems. Information network optimization problem can be interpreted as a mathematical optimization on μ-dimensional Euclidean space (tie-set flow vector space). It was shown that local optimization procedure on μ-independent tie-sets of the network led to the global optimization of the whole network. This distributed algorithm applied to max-flow problem and flow load balance problem that gave us new solving distributed methods. Proposed algorithm is essentially useful for actual information network management.