If you find a relationship between the RIS codeword and the position/s of UEs, or at least to decrease the searching overhead within the codebook, that can be possible. In my opinion, it will be something similar to positioning based beamforming in mmWave communication.

Actually, I research on such topic. So, how to model it, I am still figuring it out, back in mmWave, the beam angles can be formulated depending on the receiver direction, but now in RIS, the phase shift of elements are coupled somehow, which making finding a relation is a hard direction.

By conceptualizing channel estimation and passive beamforming in RIS-aided wireless networks as location and distribution problems within the realm of operations research, one can employ various optimization techniques, such as integer programming, convex optimization, or iterative algorithms, to obtain the optimal solution.

However, the field of channel estimation and beamforming in RIS-assisted systems is fiercely competitive, with numerous effective approaches having already been developed. As such, introducing a new algorithm-level improvement may not yield significant performance gains.

However, RIS technology is still in its early stages, harboring numerous unexplored facets. For instance, it is vital to explore the tradeoff between pilot signaling overhead and achievable performance since there exists an optimal pilot overhead that maximizes the effective achievable rate. Additionally, investigating the association of distributed RISs and multiple users as well as resource allocation and RIS optimization in wideband systems is of practical interest. These innovative scenarios offer abundant opportunities for further investigation. The knowledge of operational research might be a valuable tool in addressing these challenges.