Designing Connection Oriented Networks for Multi-Domain Path Resilience

  • Authors:
  • Angelos Lenis;Vasilis Merekoulias;Vasilis Maglaris

  • Affiliations:
  • Network Management & Optimal Design Laboratory (NETMODE), School of Electrical & Computer Engineering, National Technical University of Athens (NTUA), Athens, Greece 157 73;Network Management & Optimal Design Laboratory (NETMODE), School of Electrical & Computer Engineering, National Technical University of Athens (NTUA), Athens, Greece 157 73;Network Management & Optimal Design Laboratory (NETMODE), School of Electrical & Computer Engineering, National Technical University of Athens (NTUA), Athens, Greece 157 73

  • Venue:
  • Journal of Network and Systems Management
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

A challenge in network management and control is the ability to account for multi-domain requirements in the network planning process. Especially in Connection Oriented Networks where multi-domain path finding is a critical issue, networks need to be designed in such a manner as to restrict the possibility of erroneous inter-domain path selection. In this paper we propose per-domain topology design considerations that would leverage accurate disjoint path computations in a multi-domain environment, covering requirements of end-to-end path resiliency. In multi-domain environments state information between domains is heavily aggregated, hiding internal topology details dictated by scalability concerns, but also by restrictive domain administration policies for privacy, and security. Disjoint path finding is strongly affected by the aggregation techniques, since they do not provide information on path overlap. To handle this issue we introduce a metric, the Overlap Factor (OF), that quantifies path overlap in domains. The OF can be passed as an additional parameter of the inter-domain information exchange model to evaluate disjoint end-to-end paths. Alternatively, if domains were appropriately designed, this additional parameter might not be needed in evaluating resilient pairs of inter-domain paths. We based our recommended topology design algorithm on exploiting locally known OF values within the context of Genetic Algorithms. Extensive simulations confirm that domains designed using our proposed algorithm, result into accurate multi-domain disjoint path identification, with a high success ratio compared to networks that are designed without inter-domain considerations.