On the use of second level adaptation for networked adaptive systems

The quality of any control design can be judged by the resulting transient response. When an adaptive control law is used for linear time-invariant systems with unknown parameters, the resulting transients can be rather unsatisfactory. This problem persists when the adaptive controller receives information about the output of the plant through a wireless network. In this paper, we use the multiple model technique to improve the transient performance in this scenario. We demonstrate that the output of the plant tracks a specified reference trajectory despite packet loss and packet delay. We investigate and compare the multiple models, switching and tuning technique with the multiple model approach that uses second level adaptation. With fewer models, the latter technique uses all the information in these models when compared to the former method.