大规模通信网络带宽优化分配问题研究

提出一种大规模通信网络带宽分配的新方法．作者将大系统理论中的分解—协调方法运用于解决大规模通信网络带宽分配的优化问题，大型网络的优化问题被分解成一些互相关联的小型子网的优化问题．整个带宽优化分配问题的解决分为三个阶段：分解、协调优化及合并优化．计算结果表明，与现有算法相比，分解—协调大规模带宽管理算法(DCLPBM)既保证了很高的计算精度，又降低了时间与空间复杂度．由于算法中用到的协调机制较简单，DCLPBM易于推广到分布式计算环境，从以网络来治理网络的角度看，它具有较广的前景．

On the Optimization of Path Bandwidth Allocation for Large-Scale Communication Networks

In this paper,the authors propose a new optimization method of path bandwidth allocation for large-scale telecommunication networks. By applying the decomposition-coordination method of large-scale system theory to solve the optimization problem of path bandwidth allocation for large-scale telecommunication networks,a large-scale optimization problem is converted into small and inter-connected sub-problems. The path bandwidth allocation procedure is composed of three phases. Firstly,the large-scale network is divided into small sub-networks; then,a decomposition-coordination phase is imposed to allocate path bandwidths between nodes within each sub-network and to find the bottleneck subnet of the whole network and to evaluate its worst Call Blocking Probability (CBP); finally,path bandwidths of source-destination pairs between different sub-networks are brought out by successive application of an traditional optimization algorithm to path bandwidth allocation for source-destination pairs in sub-networks merged from adjacent subnets. Comparing with existing algorithms,results in this paper show that the Decomposition-Coordination Large-scale Path Bandwidth Management (DCLPBM) algorithm can get results with high accuracy and both spatial and time complexities are reduced. As the coordination mechanism in our algorithm is quite simple,the DCLPBM can be converted to a distributed computing version easily. This is promising from a point of view of managing the network in a real network environment.