MIMO Signal Detection Using Relaxed LLL Lattice Reduction

The patent applies parameter changes by modifying the LLL algorithm's reduction factor parameter δ. Instead of using the conventional fixed value δ=3/4, the patent dynamically adjusts δ to be less than 3/4 (0 < δ < 3/4), thereby changing the algorithm's behavior to reduce basis vector swapping operations while maintaining adequate detection performance. This parameter modification directly addresses the contradiction by lowering computational complexity without completely sacrificing detection accuracy.

The patent applies partial action by performing a reduced version of lattice reduction that doesn't fully satisfy the conventional LLL conditions. By using a relaxed reduction factor δ < 3/4, the algorithm performs only partial basis vector swapping operations needed to achieve adequate reduction, rather than exhaustively applying full LLL reduction. This partial action maintains sufficient detection accuracy while significantly reducing the time required for each reduction operation, thereby lowering processing delay.

The patent applies local quality by making the lattice reduction process adaptive to local conditions rather than applying uniform full reduction. The relaxed parameter δ < 3/4 allows the algorithm to achieve sufficient local quality in critical areas (maintaining detection performance) while avoiding excessive reduction operations in areas where they are not needed. This selective approach improves lattice basis quality where necessary while reducing overall computational overhead.