Two different research areas are connected in this paper: mobile systems and distributed deadlock resolution. As is stated in the paper, the fundamental building blocks of many traditional distributed algorithms rely on assumptions, such as message passing and static network topology, which are no longer valid if we consider that clients and servers can move and fail.
The paper focuses on a description of a distributed deadlock detection algorithm in mobile systems, and describes measurements obtained. The paper considers a single request model for the computation model, though the authors do not state this explicitly. The authors do not provide a formal description of the algorithm. A description of the algorithm, using pseudocode or a formal tool, would be very useful in completely understanding the algorithm. This would also facilitate a correctness proof to ensure that the algorithm satisfies the safety and liveness conditions that must be verified by every deadlock resolution algorithm.
This is a very interesting paper. It opens a channel to the study and analysis of distributed algorithms in more realistic environments, where movement and failures are two important characteristics that must be taken into account.