Asymmetric Attribute Aggregation in Hierarchical Networks

01/2007; DOI: 10.1093/ietcom/e90-b.8.2034
Source: OAI

ABSTRACT To achieve scalability and security, large networks are often structured hierarchically as a collection of domains. In hierarchical networks, the topology and QoS parameters of a domain have to be first aggregated before being propagated to other domains. However, topology aggregation may distort useful information. Although spanning tree aggregation can perfectly encode attribute information of symmetric networks, it can not be applied to asymmetric networks directly. In this paper, we propose a spanning tree based attribute aggregation method for asymmetric networks. The time complexity of the proposed method and the space complexity of its resulted aggregated topology are the same with that of the spanning tree aggregation method in symmetric networks. This method can guarantee that the attributes of more than half of the links in the networks are unaltered after aggregation. Simulation results show that the proposed method achieves the best tradeoff between information accuracy and space complexity among the existing asymmetric attribute aggregation methods.

  • [Show abstract] [Hide abstract]
    ABSTRACT: In an optical network, the connections are generally bidirectional, but their QoS parameters in each direction may be not the same. In this paper, we propose an enhanced algorithm called Node Label Order First (NLOF), which can maintain asymmetrical information and guarantee availability of the compressed topology. Besides, a decoding algorithm to restore the compressed topology named Average Proportional point (AP) is also proposed, which not only retains the space complexity of the aggregation process but also improves the accuracy of the restored information. Simulation results show that combing NLOF with AP can balance the contradiction between space complexity of the aggregation algorithm and routing accuracy. KeywordsMulti-domain–Topology aggregation–Asymmetric–Direction–Decoding
    Photonic Network Communication 02/2011; 21(1):28-33. · 0.45 Impact Factor