On MIMO K-Best Sphere Detector architecture complexity reductions

This paper describes a way to modify the K-Best Sphere Detector, using the knowledge that the real and the imaginary part of complex number are always orthogonal. To ensure that this property is used, a modification to the Real Value Decomposition is proposed that will ensure that the orthogonality is seen also after the decomposition. The proposed method allows for equally efficient implementations of real-valued K-Best Sphere Detector implementations as the complex-valued implementation. The cost of the lowered complexity is an increased Bit Error Rate. With proper coding this drawback could be alleviated.