Research On Linear Decoders And Their Performance In A Large-scale Antennas System

A wireless communication system with a high transmission rate, such as, a Multiple-Input,Multiple-Output (MIMO) system, is urgently required to satisfy high-speed transmissionrequirement of various applications. However, due to complexity of the radio channel and someunknown interference seriously impede the correct reception of the signal, so the decodingperformance of receiver is directly related to the quality of the communication system. Themaximum-likelihood decoder would be best decoding method, but for MIMO systems with a largenumber of antennas, this decoder has a very high computational complexity, the application is verylimited, so find a decoder which has a low decoding complexity and better decoding performancebecomes particularly important. Thus, linear receivers, such as Zero-Forcing (ZF) receiver (ordecoder) or Minimal Mean-Square Error (MMSE) receive (or decoder), have attracted people’sattention. Although these receivers are classical, and a lot of investigations have been conducted, itis still interested to find out various properties of these receivers in new environments of theirapplications. Recently, works on performance analysis of ZF or MMSE receiver have been found inliteratures.For an uplink of single cell multiuser (MU) MIMO system with M receive antennas at basestation and one transmit antenna at each user, in this paper, ZF and MMSE decoders and theirperformances are first investigated. For such a system, analysis of existing research mostlyconcentrated in the asymptotic properties. This paper gives analytical formulas of upper bounds,lower bound and approximations of pair-wise error probabilities (PEP) of ML, ZF and MMSEdecoders for given parameters M, K (number of users), and (Signal-to-Noise Ratio, SNR).These conclusions which derived show that, in the single cell system, the diversity order of both theZF decoder and the MMSE decoder is all equal to M K1, while the performance of MMSEdecoder is slightly better than the ZF decoder’s under a wild condition. In this paper, a formulaestimating MMSE decoder dB gain over ZF decoder is given. At last, simulations to confirm theseconclusions are shown.For the uplink of multi-cell multiuser (MU) MIMO system which has L cells, each cell has abase station (BS) with M antennas and K users with single antenna, the analytic forms of upperbounds, lower bounds and approximations of PEP of ZF and MMSE decoders are derived. Theseformulas show that, for the multi-cell system, if the BS knows CSI of the users in its cell only anddoes not have CSI of the users in other cells, error floors will occur. In other words, even whenSignal-to-Noise Ratio (SNR) goes to infinite, the error probability of two decoder won’t go to zero, while the error floors disappear when M goes to large. These results encourage us to consider and investigate a massive MIMO system, where the BS equips a large number of antennas.The last conclusions of this thesis is assumed that the equivalent received SNR is scaled as M-a and M goes to infinite, then if and only if0<a<1, the error probability of ZF and MMSE decoders goes to zero. When a=1, the error floors occur, that is, error probabilities won’t go to zero, which is very different from the result shown in a paper. At last, all theoretical results above are confirmed by simulations.