Precoding techniques in downlink for massive MIMO systems

Abstract

MIMO Multiple Input Multiple Output, is a present day technology which is being applied by having multiple antennas at both transmitter and receiver end because of its excellentand improved performance in terms of data rate and bit error rate. This technology when used tocommunicate with multiple terminals simultaneously is termed as multiuser MIMO. It offersmore advantages then usual technique of point-to-point MIMO such as data rate is increasedbecause multiple streams of data are travelling and they are entertained at the receiver end at thesame time. MU MIMO also offers improved energy efficiency due to the fact that the basestation is aware of terminals locations and hence can focus its emitted energy in spatialdirections. Moreover interference factor is reduced because the base station deliberately avoidsusing those paths for transmission which possess harmful interference. Massive MIMO having array of hundreds of antennas serve many tens of terminals at the same time utilizing single time-frequency resource. The basic objective of massive MIMO ishaving all the advantages of conventional MIMO on much larger scale. In Massive MIMO system, adapting the transmitted signal across multiple users gives extra advantages but withincreased number of antennas both at transmitter and receiver side, level of channel knowledgerequired increases proportionally. This creates challenging problem in practical systems as lot offeedback is sent back to the transmitter because it does not have priori channel information. The main objective of this thesis is to devise a technique that can feedback largeinformation about channel back to the transmitter optimally. Feedback having channel stateinformation (CSI) is required at the transmitter to use it effectively for pre-coding. Channelestimations with the help of Minimum Mean Square Error (MMSE), Least Square (LS) andDiscrete Fourier Transform (DFT) with LS and MMSE are carried out for 2-tap channels indownlink (DL). Complete channel information for all channels is gathered and CompressSensing (CS) Algorithm is applied in the reverse channel to minimize the number of bitsrequired to send the information back to transmitter effectively. Impact of truncation of channelestimates on total feedback bits is studied which shows that comparative results can be attainedby reducing significant figures hence decreasing number of bits.