Comparing LQG/LTR and the SDRE techniques for hybrid fully-connected PLL network control

• Autor(a) principal: Bueno, Atila Madureira [UNESP]
• Data de Publicação: 2013
• Outros Autores: Tusset, Angelo Marcelo, Correa, Diego Paolo Ferruzzo, Piqueira, José Roberto Castilho, Balthazar, José Manoel [UNESP]
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1115/DETC2013-12649; http://hdl.handle.net/11449/227641

Synchronization plays an important role in telecommunication systems and integrated circuits. The Master-Slave is a commonly used strategy for clock signal distribution. However, due to the wireless networks development and the higher operation frequency of integrated circuits, the Mutually-Connected clock distribution strategies are becoming important, and the Fully-Connected strategy appears as a convenient engineering solution. The main drawback of the Fully-Connected architecture is the definition of control algorithms that assure the stability of the network sinchronization. In hybrid synchronization techniques groups of nodes synchronized by the Fully-Connected architecture are synchronized with network master clocks by using the Master-Slave tecnique. In this arrangement, if a route of clock signal distribution becomes inoperative, the group of Fully-Connected nodes retain for some time the original phaseand frequency received from the network. The Fully-Connected architecture complexity imposes difficulties to satisfy both stability and performance requirements in the control system design. For that reason the multi-variable LQG/LTR and the SDRE control techniques are applied in order to fulfill both stability and performance requirements. The performance of both techniques are compared, and the results seems to confirm the improvement in the transient response and in the precision of the clock distribution process. Copyright © 2013 by ASME.