中文摘要：

针对以最小化网络阻塞率为目标的光网络路由及波长分配（RWA）问题，考虑到全网结构不均衡易导致部分链路负载过高，进而造成全网阻塞率过高问题，在基于爱尔兰损失公式的链路阻塞概率模型的基础上，建立了最大化路径畅通概率的优化模型。为了克服优化模型的非线性造成的求解困难，借鉴大系统中分解协调的思想对链路负载进行预估，将原优化问题转化成乘积最长路问题，并结合负载滚动预估更新及类Dijkstra算法进行近似求解。仿真比较实验表明，本文算法能够较好地近似求解所提出的最大化畅通概率模型，有效地均衡了全网负载，降低了全网阻塞率，提高了网络传输性能。

英文摘要：

The key problem of wavelength division multiplexing （WDM） optical network is the routing and wavelength assignment （RWA） problem which has the objective to minimize the blocking probabili- ty of the whole network. But the phenomenon of heavy-load over some particular links, which results from the unbalance of the network structure, may bring to high blocking probability. Therefore a link blocking probability model is proposed based on the Erlang loss formula in this paper. Then an optimiza- tion model is established to maximize the route unblocking probability. The idea of decomposition coordi- nation is utilized in order to overcome the nonlinearity by setting the link load and converting the original problem to a shortest path problem. Eventually,a similar Dijkstra algorithm,which is combined with the updating of link load,is designed to iteratively solve the shortest path problem. Simulation results show that the proposed algorithm can approximately solve the original optimization problem and at most re- duce blocking probability by 27.6% compared with the shortest path （SP） algorithm and LBFR algo- rithm. So the proposed algorithm can efficiently balance the load of the whole network, get the network congestion reduced and finally improve the transmission performance.