Michael Moore1, Mark Davenport
15:15 - 16:45 | Wed 6 Jul | Pentland A | R11.9
In this paper we consider the problem of characterizing and analyzing a wireless network from limited passive observations of network activity. In particular, we will assume that the only information that we can acquire is knowledge of when each particular transmitter in the network initiates any given transmission. From this data, we wish to be able to solve problems such as learning the network topology, detecting changes to the existing topology, and extracting higher-level summaries of information flow in the network. We show how one can use a multidimensional autoregressive point process known as a Hawkes process to model the observed data and approach these problems.