Agrupamento de usuários e otimização da reflexão baseados em informações estatísticas de canal para sistemas de comunicação multiusuários auxiliados por superfícies inteligentes reconfiguráveis

Abstract

Reconfigurable Intelligent Surfaces (RIS) are regarded as a key technology for enhancing the communication systems of future generations. The performance gains provided by RIS are well established in the literature; however, most authors assume full knowledge of the wireless channel, which implies a significant network training overhead. Furthermore, the typical deployment scenario assumes that the reflection coefficients are estimated at the base station and transmitted to the RIS controller via a wireless control link with limited data rate, which may not be sufficient to convey all coefficients within the channel coherence interval. The non-ideal nature of the control link imposes constraints on the RIS update rate, as confirmed by prototy- pes whose reconfiguration times are non-negligible and must be considered to ensure system synchronization. A promising scenario under these imposed limitations involves time-division multiple access (TDMA) techniques, whereby a fixed configuration of the reflection coefficients is applied to a given set of temporally multiplexed users, along with coefficients based on statistical channel state information, which remain valid over longer time and frequency intervals. In this context, a heuristic method for user clustering and phase shift design is proposed, relying solely on users' requirements and using statistical channel state information. For a scenario in which each user is served exclusively in a TDMA slot, with no multiuser interference, the performance results indicate the feasibility of the proposed algorithm. When the methodology is extended to scenarios with multiple users served simultaneously within the same TDMA slot, the performance metrics reveal a limitation in RIS operation in multiuser environments. The adopted approach was not able to adequately compensate for the interference effects between users nor for the constraints imposed by resource sharing. The results suggest that efficient use of RIS requires time resource allocation schemes to avoid the simultaneous service of multiple users.