MIMO Decoder Soft Bit Computation via Staged Distance Metrics

The patent segments the LLR computation process into distinct stages: computing minimum distances for hypotheses and separate computations for counter-hypotheses. This segmentation allows the decoder to process distance metrics in an organized manner without requiring extensive storage, as each stage produces intermediate results that are immediately used in subsequent stages, thereby reducing memory requirements while maintaining computational accuracy.

The patent performs preliminary computation of minimum distances for all hypotheses before computing distances for counter-hypotheses. This preliminary action establishes a foundation of sorted distance metrics that enables efficient LLR computation without requiring all metrics to be stored simultaneously. The preliminary sorted results are used immediately to compute LLRs, avoiding the need for extensive storage of cumulative metrics.

The patent performs preliminary sorting of distance metrics for hypotheses before computing counter-hypothesis distances. This preliminary action ensures that the most reliable metrics are identified and used first in LLR computation, improving reliability while reducing the time required for subsequent processing. By pre-sorting metrics, the system avoids repeated sorting operations and can quickly access the most relevant distance values.

The patent extracts only the necessary distance metrics from the complete set of computed distances. Instead of storing and processing all partial and cumulative distance metrics, the system extracts and uses only those metrics required for LLR computation - specifically the minimum distances for hypotheses and the necessary counter-hypothesis distances. This extraction reduces computational time while maintaining the reliability needed for accurate error-rate performance.

The patent segments the sorting operation into two distinct phases: sorting distance metrics for hypotheses and separate processing for counter-hypotheses. This segmentation allows the system to maintain measurement precision by properly sorting metrics where needed, while avoiding the complexity of sorting all metrics simultaneously. The segmented approach processes only the necessary subsets of metrics, reducing storage requirements and device complexity.

The patent performs preliminary sorting of distance metrics only for hypotheses before computing counter-hypothesis distances. This preliminary sorting ensures measurement precision for the primary hypotheses while avoiding the need to sort all possible metrics. The pre-sorted hypothesis metrics are then used efficiently in LLR computation, maintaining accuracy without requiring extensive sorting infrastructure for the complete metric set.