Revisited Design Criteria For STBCs With Reduced Complexity ML Decoding
The design of linear STBCs offering a low-complexity ML decoding using the well known Sphere Decoder (SD) has been extensively studied in last years. The first considered approach to derive design criteria for the construction of such codes is based on the Hurwitz-Radon (HR) Theory for mutual orthogonality between the weight matrices defining the linear code. This appproach served to construct new families of codes admitting fast sphere decoding such as multi-group decodable, fast decodable, and fast-group decodable codes. In a second Quadratic Form approach, the Fast Sphere Decoding (FSD) complexity of linear STBCs is captured by a Hurwitz Radon Quadratic Form (HRQF) matrix based in its essence on the HR Theory. In this work, we revisit the structure of weight matrices for STBCs to admit Fast Sphere decoding. We first propose novel sufficient conditions and design criteria for reduced-complexity ML decodable linear STBCs considering an arbitrary number of antennas and linear STBCs of an arbitrary coding rate. Then we apply the derived criteria to the three families of codes mentioned above and provide analytical proofs showing that the FSD complexity depends only on the weight matrices and their ordering and not on the channel gains or the number of antennas and explain why the so far used HR theory-based approaches are suboptimal.