Automatic network restoration using a two-level associative memory

Based on the principles of neural computing, a new approach in network restoration is proposed and implemented in this paper. This new approach, called NRNN (network restoration using neural networks), is a hybrid algorithm and is implemented by using a two-level associative memory. Compared to NETSPAR (a hybrid algorithm proposed by Bellcore in 1991), NRNN not only obtains 100% recovery of disrupted traffic due to a link or node failure but also requires a very small memory at each node. At a node, the memory requirement for NRNN is less than 1% of the memory requirement for NETSPAR.<
>