Base Station Uplink Reception With Adaptive Covariance Selection

The patent segments the noise-and-interference covariance matrix estimation by reference signal type. Instead of using a single covariance matrix for all reference signals, the system estimates separate covariance matrices for different reference signal types (e.g., first reference signals and second reference signals). This segmentation allows each matrix to be optimized for its specific reference signal characteristics, improving reception performance while managing computational complexity through targeted estimation rather than universal estimation.

The patent applies local quality by tailoring the noise-and-interference covariance matrix to specific reference signal locations and types. The system determines covariance matrices based on the specific characteristics of each reference signal (e.g., time-frequency resources, signal type), ensuring that each local region or signal type receives appropriate interference estimation. This localized approach improves reception performance for specific signals without requiring complex global estimation for all signals.

The patent performs preliminary action by pre-estimating noise-and-interference covariance matrices for different reference signal types before actual data reception. The system estimates these covariance matrices in advance based on reference signal characteristics, storing them for use during subsequent channel estimation. This preliminary estimation avoids the need for complex real-time interference calculation during data reception, improving reception performance while maintaining computational efficiency through pre-computed values.

The patent employs feedback mechanisms where the system uses received reference signals to estimate covariance matrices, which then feed back into the channel estimation process. The estimated covariance information is fed back to improve subsequent data symbol estimation, creating an iterative optimization that enhances reception performance while using efficient feedback-based updates rather than exhaustive recomputation.