Controlled Dissemination Filter (CDF) for Integrated Link Selection Agents

Abstract

To perform optimizations as the network changes dynamically, distributed network agents rely on information that is dispersed in the network. A data dissemination service for the agents can benefit from data filtering at selected nodes in a data distribution tree. Controlled dissemination filter (CDF) offers an efficient filtering functionality such as sifting, collection, aggregation, and scoping. We defined a number of new objective functions for the data-flow channelization algorithms, and created both centralized and distributed algorithms to optimize filter placement in polynomial time. We also designed an application of the CDF to information flows relevant to the unicast routing control agent (URCA) and showed how the CDF reduces the bandwidth consumed by data dissemination without impairing the URCA decision-making functionalities.