Codebook Information Processing for 5G PMI Selection

The patent segments the PMI selection process into three independent levels: first PMI1 selection, first PMI2 selection, and second PMI selection. Each level searches a reduced subset of codebooks rather than traversing all codebooks simultaneously. This segmentation reduces the search space from O(N1×N2×N3) to O(N1+N2+N3), significantly lowering calculation complexity while maintaining feedback accuracy.

The patent transforms the traditional two-dimensional codebook structure (horizontal and vertical beam directions) into a three-dimensional search space by introducing the second PMI for subband-level adjustments. This dimensional expansion allows the system to achieve more precise channel state information feedback by searching across multiple dimensions independently, rather than traversing the entire combined space.

The patent divides the large codebook into three manageable segments corresponding to first PMI1, first PMI2, and second PMI. Each segment represents a specific aspect of precoding (wideband horizontal, wideband vertical, and subband adjustments respectively). This segmentation allows the system to support a large number of codebooks for high-resolution precoding while reducing the computational burden by searching each segment independently rather than traversing all codebooks at once.

The patent introduces dynamic adaptation by allowing different search strategies and parameters for each PMI level. The first PMI1 and PMI2 operate at wideband level providing stable baseline precoding, while the second PMI operates at subband level providing dynamic adjustments. This dynamic multi-level approach enables the system to adapt to varying channel conditions without requiring exhaustive search across all codebooks.