Mimo Detection Of Vlsi Architecture Research

Multiple-input multiple-output (MIMO) is a key technology for future wirelesscommunication systems. In particular the use of spatial multiplexing allows achievinghigher capacity and spectral efficiency. The capacity will be increased linearly with theminimum of the number of antennas employed at the transmitter and at the receiver.However, capacity and spectral efficiency gains cost at the expense of highercomputation complexity and silicon complexity, bringing negative impacts on both thereceiver’s throughput and power consumptions.This thesis demonstrates some widely recognized MIMO detection algorithms,shows simulation results for choosing proper algorithms under a series of scenarios andgives close-to RTL architectures using sphere decoding algorithm.This thesis discusses hard/soft-output maximum likelihood detection algorithms,hard/soft-output sphere decoding and hard/soft-output fixed-complexity spheredecoding. It also discusses channel preprocessing algorithms for sphere decoding andthat for fixed-complexity sphere decoding. A method has been proposed to address theBER performance degradation problem caused by LLR enlarging distortion which isprobably caused by losing one or more counter-hypothesis paths.Architecture of maximum likelihood algorithm is reviewed firstly. Then,architecture of sphere decoding is not only reviewed, but is improved and detailednearly into RTL level. Works include: combining redundant parts, re-organizing toparchitecture, giving detailed definition of sub modules for RTL development.Finally, algorithms are compared through a series of simulation scenarios. Basedon the comparison of results, proper algorithms are selected.