Performance Analysis And Receiver Design In Multiple Antenna Systems With Phase Noise

To meet the challenge of significant and continuing growth of mobile traffic and wireless-connected de-vices, the industry has embarked defining and standardizing of the 5th generation (5G) mobile communication system. How to improve the spectral efficiency and power efficiency of wireless mobile communications is the main task of 5G, as a result, there is a need for new system architecture and therefore the design of com-prehensive and high capacity backhauls becomes a significant factor in 5G system. The use of high-order constellations to increase throughput on backhauls makes the overall system extremely sensitive to phase noise, this problem has stimulated a large body of research on the the impact of phase noise on the throughput of backhaul links. In this thesis, the performance and the receiver design of multiple-antenna systems affected by phase noise will be investigated.Firstly, we study the performance of the mutiple antenna system affected by common phase noise and derive the capacity expressions. Specifically, we consider the scenario where a single oscillator drives all radio frequency circuitries at each transceiver and the analyses incorporate both unitary and non-unitary channel matrix. For unitary channel matrix scenario, we use mutual information and entropy to present the non-asymptotic capacity upper bound and the asymptotic capacity characterization; when the channel matrix is non-unitary, we split the M × M multiple input multiple output (MIMO) system into a single input single output (SISO) system and a M-1 × M-1 MIMO system, the SISO system is used to estimate the current phase noise while transmitting signals, then the estimated value is applied in the M 1 × M -1 MIMO system to get rid of the phase noise, in this way, the second part of the system is simplified into an additive white Gaussian noise (AWGN) system and its capacity can be accurately obtained by water-filling strategy, by doing this we can get a tight capacity bound for non-unitary channel case. We compare our bounds with the lower bounds obtained by evaluating numerically the information rates achievable with quadrature amplitude modulation (QAM) constellations using the factor graph algorithm, simulation results show that, our bounds can describe accurately the capacity behavior of the system.Subsequently, the multiple antenna systems impaired by independent phase noise processes are inves-tigated and the capacity expressions under different antenna deployment are derived. We first consider the single input multiple output (SIMO) and the multiple input single output (MISO) systems, we provide capac-ity bounds for both systems as well as the discussions on practical issues such as how to achieve the system capacities. Then the MIMO systems driven by separate oscillators are considered, we propose a decompo-sition method, where the MIMO system is decomposed into MISO systems, one MISO system is first used to estimate the transmit phase noise samples, the estimated values are then applied to eliminate the trans-mit phase noise in other MISO systems so that the remaining MISO systems are only affected by their own receive phase noise. Simulation results show that, we obtain tight MIMO capacity bounds by applying the above strategy.Then, we focus on the multiplexing gain of 2 x 2 MIMO system impaired by independent phase noises.The characterization is obtained by proving that the lower bound matches the upper bound. The lower bound is obtained by achievability proof, that is, to present a multiplexing gain which can be achieved in real sys-tems. To prove the converse part, we use the duality technique and carefully choose the multivariate gamma distribution in the duality approach to get a tight upper bound of the multiplexing gain. The result shows that no MIMO multiplexing gain is to be expected when the phase noise processes at each antenna are indepen-dent, in other words, extra transmit and receive antennas do not provide additional degrees of freedom in the presence of independent phase noises, this result indicates distinct differences between phase noise systems and AWGN systems.Finally, we address the problem of maximum likelihood (ML) receiver in the presence of random phase noise and come up with a simple ML decision rule which only involves the conditional mean and variance of the phase noise. We first perform Taylor expansion to the original expression and show that the ML detec-tor for this problem can be formulated as a weighted sum of central moments of the conditional probability density function (PDF) of phase noise. We then propose a simplified decision rule in light of the computa-tional complexity in practice. The approximate algorithm is obtained by truncating the original decision rule, specifically, we truncate the ML rule to two terms for detection so that it is only related to the conditional mean and variance of the phase noise. We also give the upper bound on the error for the above-mentioned approximation with respect to the original ML rule and develop insights on how the decision performance is influenced by system parameters. Simulation results demonstrate that under the simplified decision rule, low symbol error probability (SEP) performance can be guaranteed without high computational complexity, which present its efficiency and low complexity.