I have modified your code in order to implement a method that associates users with the available RISs, allocates the transmission power of multiple users and configures the phase shifts of multiple RISs. The state now is defined as the achievable SNR per user, however the action space becomes very large when the number N of RIS elements (per RIS) is very high (e.g., 64 or 128) or more users and RISs are added. The complexity is very high, any hints on how this could be improved algorithm-wise?
I have modified your code in order to implement a method that associates users with the available RISs, allocates the transmission power of multiple users and configures the phase shifts of multiple RISs. The state now is defined as the achievable SNR per user, however the action space becomes very large when the number N of RIS elements (per RIS) is very high (e.g., 64 or 128) or more users and RISs are added. The complexity is very high, any hints on how this could be improved algorithm-wise?