|dc.description.abstract||Wireline backbone networks have evolved to support many different end-user bandwidth services, including private leased lines, packet voice, high-definition packet video, virtual private networks (VPN), etc. From a functional perspective, these networks are usually comprised of different technology layers and segmented into multiple domains, each managed by its own authority and operating under its own policy.
However, as business/mission-critical demands and stringencies continue to grow, there is a pressing need to improve network/service survivability (reliability) across multiple
operating domains. Indeed, this is a very challenging problem owing to the obvious scalability and confidentiality limitations across domain boundaries. As a result, multidomain survivability schemes must be able to operate in a distributed, decentralized
manner. In addition, new recovery provisions are also needed to handle large-scale “catastrophic” events causing multiple correlated failures, e.g., natural disasters, power outages, and weapons of mass destruction (WMD) attacks.
Overall, a wide range of survivability solutions have been developed over the years, albeit mostly focusing on single domain settings. Broadly, these schemes can be classified as pre-provisioned protection and post-fault restoration type strategies.
Namely, the former types offer fast recovery and single-failure resiliency guarantees via
the pre-allocation of backup resources. Meanwhile, the latter offer more latent nondeterministic
recovery via post-fault path re-routing. In general, both of these strategies can be extended across domain boundaries to achieve multi-domain service recovery. However, most existing proposals in this space only study protection-based approaches and do not incorporate end-to-end resource efficiency or backup diversity concerns.
Further treatments for multi-failure recovery are also lacking.
In light of the above, this dissertation proposes a range of solutions to improve distributed multi-domain survivability. First, a new protection scheme is proposed to achieve improved end-to-end path pair diversity (resiliency) with reduced routing
overheads. This solution is then augmented with probabilistic considerations to handle more challenging multi-failure scenarios, i.e., as arising during catastrophic events. In particular, this solution incorporates both resource efficiency and failure risk mitigation concerns. Finally, a novel post-fault restoration scheme is proposed to achieve multidomain failure recovery using joint intra-/inter-domain crankback signaling. The
performance of all of these proposed solutions is also evaluated using discrete event
simulation and appropriate comparisons made versus existing strategies.||en_US