MIMO Signal Detection via QR Decomposition and Zero-Forcing

The patent segments the detection process into two distinct stages: (1) QR decomposition stage that processes the channel matrix to obtain triangularized forms, and (2) detection stage that performs simplified calculations using the preprocessed matrices. This segmentation allows the computationally intensive matrix decomposition to be performed once, followed by efficient detection calculations, thereby reducing overall complexity while maintaining accuracy.

The patent applies preliminary action by performing QR decomposition of the channel matrix before the actual detection process. The channel matrix is preprocessed into triangular forms (R1 and R2) and the received signal is pre-multiplied by QH to obtain transformed signals. This preliminary processing simplifies subsequent detection calculations, avoiding the need for exhaustive searches during the actual detection phase.

The detection algorithm is segmented into preprocessing (QR decomposition) and detection phases. For high-order constellations, the QR decomposition is performed once regardless of constellation size, and the simplified detection phase efficiently handles the increased number of constellation points without proportionally increasing processing time, thus enabling high throughput with manageable processing delays.

The patent changes the parameter representation by transforming the detection problem into a simplified form using QR decomposition. The original complex detection problem with multiple intertwined signals is transformed into a triangularized form where detection can proceed sequentially with reduced computational burden, allowing high-order constellations to be processed efficiently.

The patent performs preliminary QR decomposition to transform the channel matrix into triangular forms before detection. This preliminary action simplifies the detection calculations, reducing the number of complex matrix operations required during actual detection. The simplified detection process using transformed signals and triangular matrices reduces computational load, thereby lowering power consumption while maintaining detection accuracy.

The patent extracts the computationally intensive matrix decomposition operations and performs them separately as a preprocessing step. The main detection process then uses only simplified calculations based on the extracted triangular forms, reducing the computational burden and power consumption during the critical detection phase while preserving accuracy.