An economic analysis of routing conflict and its resolution

Abstract

Thousands of competing autonomous systems (ASes) must cooperate with each other to provide global Internet connectivity. Each AS has independent economic objectives and retains autonomy in setting their routing policies independently to maximize its profit. However, such autonomy enables ASes to produce conflicting routing polices and thus raises route oscillations between them (i.e., routing divergence). This paper studies the basic problem of routing divergence by investigating real ISP pricing data. We first demonstrate that routing divergences occur under economic dependency cycles, i.e., provider–customer cycles, of different ASes which are raised by economic conflicts between themselves. We then propose a provable cycle-breaking routing mechanism to detect and solve economic conflicts and route divergence. We show that every cycle-breaking strategy allows ASes to maximize their own profits to converge to a Nash equilibrium with a profit-sharing mechanism derived from the coalition game concept of <i>Shapely value</i>. At the Nash equilibrium point, the cycle-breaking strategies maximize ASes’ profits and encourage ASes so as to ensure divergence-free routing.