The paper deals with the stabilization of nonlinear
systems in which the loop is closed over a lossy nonacknowledged
communication network. Given a Regional Inputto-
State (ISS) stabilizing state-feedback control law, designed
without accounting for the network-induced delays, we propose
a non-acknowledged communication policy that allows to deploy
the above controller over the network without any modification,
while preserving the Regional ISS property. The time-varying
delays and packet dropouts occurring on both the up-link and
the down-link are compensated through a model-based prediction
scheme and a packet-management policy based on time-stamping.
The consistency of the prediction, which is a major issue in
the context of nonlinear systems with an embedded networked
controller, is guaranteed through the exploitation of a novel moveblocking
strategy for computing the command sequence to be
forwarded to the actuators.
