It was demonstrated in transport networks that\udnetwork (diversity) coding can provide sufficient redundancy to\udensure instantaneous single link failure recovery, while nearoptimal\udbandwidth efficiency can be reached. However, in the\udresulting multi-path routing problem the end-to-end delays were\udnot considered. On the other hand, even in a European-scale\udnetwork the delay difference of the paths has severe effect on\udthe Quality-of-Service of application scenarios, such as video\udstreaming. Thus, in this paper we thoroughly investigate survivable\udrouting in Software Defined Networks (SDNs) with several\udadditional delay bounds to the bandwidth cost minimization\udproblem. We build on the fact that, if the user data is split into\udat most two parts, then the minimum cost coding solution has\uda well-defined acyclic structure of subsequent paths and disjoint\udpath-pairs between the communication end-points. Complexity\udanalysis and integer linear programs are provided to solve these\uddelay aware survivable routing problems in SDNs.
展开▼