Fixed effort sphere decoder for MIMO OFDM systems

Abstract

A new fixed effort Multiple Input Multiple Output (MIMO) decoder for orthogonal frequency division multiplexing (OFDM) is proposed. For enhancing the data rates in wireless communication systems a higher bandwidth is needed, resulting in a frequency selective channel with multiple reflecting transmission paths. These requirements implicate additional considerations on the dimension of channel memory. Therefore we get a higher complexity for the decoder on the receiver side that can be reduced by performing the equalization in the frequency domain with OFDM, where each subchannel can be considered as a flat MIMO channel. To achieve the optimum hardware exploitation it is strongly necessary to find optimized effort based solutions to be able to implement an efficient MIMO decoder in concrete practical systems. Therefore our reduced search sphere decoder (RS-SD) algorithm that performs the closest point search over a reduced set of candidates based on pre-computed search set missing probabilities and other common fixed effort algorithms for flat MIMO channels are adapted to the OFDM system model. Furthermore our new OFDM-RS-SD algorithm is proposed quartering the needed computational effort compared to the RS-SD by additionally considering the conditions of the different subchannel matrices during the selection of the candidates. Therefore the OFDM-RS-SD offers a higher flexibility reaching near ML performance with significantly reduced computational effort.