Optimal Routing with Single Backup Path Protection

msra(2007)

引用 33|浏览3
暂无评分
摘要
Communication networks are increasingly required to be reliable. In this paper we only consider single cable failures (7, 12). The classic path protection method employed in circuit switched networks is 1+1 protection, where two cable disjoint circuits (and hence arc disjoint circuits) are established and actively used. In case an arc fails on one path, the other path will survive and enable the receiving node to restore communication by just switching to the other incoming signal. This method is simple, there are well defined standards, but the required network capacity is always more than twice the required Non-Failure (NF) network capacity. In this paper we will consider traffic engineering optimization methods for the Single Backup Path Protection (SBPP) method for circuit switched networks. This protection method is also called Shared Backup Path Protection (7) or Global Backup Path Protection (3). The SBPP method is a slight variation of 1+1 protection: Instead of actively sending data packets on both paths, one path is designated the primary path — and only when the primary path fails are data packets sent along the other so-called backup path. Since we only guarantee protection against single cable failures, several backup paths may share capacity in the network. In order to compare various protection methods, we define the following two measures: 1) Restoration Over Build (ROB) network capacity, which is the extra network capacity necessary to ensure protection, i.e. the network capacity subtracted by NF network capacity, assuming shortest path routing. 2) Relative Restora- tion Over Build (RROB) network capacity, which is the relative extra network capacity necessary to ensure protection, i.e. ROB network capacity divided by NF network capacity. In order to utilize the path protection methods traffic engineering has to be performed in order to minimize the required RROB network capacity. When working with 1+1 protection this is a well studied problem for
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要