Self-Adjusting Linear Networks with Ladder Demand Graph

Abstract

This paper revisits the problem of designing online algorithms for self-adjusting linear networks which dynamically adapt to a traffic pattern they serve. We refer to the graph formed by the pairwise communication requests as the demand graph. Even though the line is a fundamental network topology, existing results only study linear demand graphs. In this work, we take a first step toward studying more general demand graphs. We present a self-adjusting algorithm that responds to the traffic pattern drawn from the n-ladder graphThe resulting algorithm appears to have optimal competitive ratio. As two additional side results, we get a generic algorithm for an arbitrary demand graph and an optimal algorithm for a cycle demand graph.