Near-optimal mapping of network states using probes

Abstract

In many applications, such as the Internet and in-frastructure networks, nodes fail or get congested dy-namically. We study the problem of inferring all the failed nodes, when only a sample of the failures is known, and there exist correlations between node fail-ures/congestion in networks. We formalize this as the GraphStateInf problem, using the Minimum De-scription Length (MDL) principle. We propose the GraphMap algorithm for minimizing the MDL cost, and show that it gives an additive approximation, rela-tive to the optimal. We evaluate our methods on syn-thetic and real datasets, which includes one fromWAZE which gives traffic incident reports for the city of Boston. We find that our method gives promising results in re-covering the missing failures.