Event-Triggered Control over Sensor/Actuator Networks
Speaker: Manuel Mazo Jr. (UCLA)
Advisor: Paulo Tabuada 

Abstract
Wireless networks have been extensively studied as the sensing infrastructure for the most diverse applications. Less attention has been payed to the use of wireless networks as the infrastructure for sensing and actuation. Nevertheless, there are several applications that would benefit from this new paradigm such as manufacturing plants, sewer and water distribution and management networks, or precision agriculture. Performing actuation when the feedback and command signals are transmitted over wireless connections poses unique problems. On the one hand, we would like to transmit measurements as often as possible in order to react to changes in the physical environment, on the other hand, transmitting measurements too often may significantly reduce the life time of the nodes in the sensor/actuator network. We present a novel event-based technique to perform control over sensor/actuator networks, one that reduces the number of transmissions from the sensors while still enforcing desired levels of control performance. Instead of relying on periodic sensor transmissions, as is typical for traditional control applications, we propose the use of event-triggered sampling. We introduce a general framework for event-triggered control over sensor/actuator networks and prove that it guarantees the required control performance. We also show through numerical simulations that it leads to less transmissions when compared to classic periodic control.