Abstract:
An embodiment of a method and device for detecting a signal and generating bit soft-output of a multiple-input multiple-output system is provided. The device includes at least one channel estimates pre-processing unit, one received vector processing and one detection and soft-output generation unit. The pre-processing unit calculates multiple QR Decompositions of the input channel estimation matrix. The detection and soft-output generation unit computes near optimal bit soft output information with a deterministic complexity and latency. It may implement a reduced complexity search method. Globally, embodiments of the invention may allow achieving low complexity, high data rate, scalability in terms of the dimension of the MIMO system and flexibility versus the supported modulation order, all potentially key factors for most MIMO wireless transmission applications.
Abstract:
One or more techniques to iteratively detect (320) and decode data transmitted in a wireless ( e . g ., MIMO-OFDM) communication system, featuring a MIMO detector and a soft input soft-output (SISO) error-correction-code (ECC) decoder. More specifically, a method suitable for iterative detection and decoding schemes is proposed, which is able to output near optimal bit soft information ( e . g ., log-likelihood ratios, LLRs) processing efficiently given input bit soft information. First, a transmitting source is selected as a reference layer, wherein the associated symbol represents a reference transmit symbol. Subsequently, a set of candidate values are identified for the reference transmit symbol. For each candidate value a candidate transmit sequence ( U t ( X )) is estimated (616) through a novel spatial decision feedback equalization (DFE) process based on both Euclidean distance metrics and the a-priori soft information ( L a ) provided by the SISO ECC decoder. The novel DFE technique uses a novel bit metric. Techniques are provided to identify a reduced size transmit symbol candidate set and generate from it near-optimal LLRs, also processing input a-priori LLRs in an iterative fashion.