The basic idea of the paper is to look at whether an attacker who measures vehicle motions locally can estimate the full network’s state. In a double-integrator-network (DIN) model there are nodes with sensors that take local measurements, communicate locally, and then adjust their own state (for example, speed) accordingly. The authors look at two cases: (1) when observations are not noisy and (2) when the adversary’s observations are noisy. The authors consider a 1D case where each vehicle only moves in one dimension. Earlier work referenced in the paper shows that the multi-dimensional case can be transformed into a single dimension with more vehicles. The paper is well structured but lacks an in-depth introduction to the topic, so unless the reader is familiar with the general topic, she or he will have a hard time following its main sections.

The paper is an extended version of a conference paper presented at the 51st IEEE Conference on Decision and Control (CDC). However, it is still only a brief paper with about five pages of text. For the nonexpert reader, it is hard to understand the overall goal of the research and the results because the introduction is very short and there is no conclusion or summary at the end of the paper. The authors’ desire to present all of their information in a brief paper is understandable, but it seems that the paper’s short length forced the authors to cut all general parts.