Future cellular systems: fundamentals and the role of large antenna arrays
MetadataShow full item record
In this thesis, we analyze the performance of three promising technologies being considered for future fifth generation (5G) and beyond wireless communication systems, with primary goals to: i) render 10-100 times higher user data rate, ii) serve 10-100 times more users simultaneously, iii) 1000 times more data volume per unit area, iv) improve energy efficiency on the order of 100 times, and iv) provide higher bandwidths. Accordingly, we focus on massive multiple-input multiple-output (MIMO) systems and other future wireless technologies, namely millimeter wave (mmWave) and full-duplex (FD) systems that are being considered to fulfill the above requirements. We begin by focusing on fundamental performance limits of massive MIMO systems under practical constraints such as low complexity processing, array size and limited physical space. First, we analyze the performance of a massive MIMO base station (BS) serving spatially distributed multi-antenna users within a fixed coverage area. Stochastic geometry is used to characterize the spatially distributed users while large dimensional random matrix theory is used to achieve deterministic approximations of the sum rate of the system. We then examine the deployment of a massive MIMO BS and the resulting energy efficiency (EE) by considering a more realistic set-up of a rectangular array with increasing antenna elements within a fixed physical space. The effects of mutual coupling and correlation among the BS antennas are incorporated by deriving a practical mutual coupling matrix which considers coupling among all antenna elements within the BS. Accordingly, the optimum number of antennas that can be deployed for a particular antenna spacing when EE is considered as a design criteria is derived. Also, it is found that mutual coupling effect reduces the EE of the massive system by around 40-45% depending on the precoder/receiver used and the physical space available for antenna deployment. After establishing the constraints of antenna spacing on massive MIMO systems for the current microwave spectrum, we shift our focus to mmWave frequencies (more than 100GHz available bandwidth), where the wavelength is very small and as a result more antennas can be rigged within a constrained space. Accordingly, we integrate the massive MIMO technology with mmWave networks. In particular, we analyze the performance of a mmWave network consisting of spatially distributed BS equipped with very large uniform circular arrays (UCA) serving spatially distributed users within a fixed coverage area. The use of UCA is due to its capability of scanning through both the azimuth as well as elevation dimensions. We show that using such 3D massive MIMO techniques in mmWave systems yield significant performance gains. Further, we show the effect of blockages and path loss on mmWave networks. Since blockages are found to be quite detrimental to mmWave networks, we create alternative propagation paths with the aid of relays. In particular, we consider the deployment of relays in outdoor mmWave networks and then derive expressions for the coverage probability and transmission capacity from sources to a destination for such relay aided mmWave networks using stochastic geometric tools. Overall, relay aided mmWave transmission is seen to improve the signal to noise ratio at the destination by around 5-10dB with respect to specific coverage probabilities. Finally, due to the fact that the current half duplex (HD) mode transmission only utilizes half the spectrum at the same time in the same frequency, we consider a multiuser MIMO cellular system, where a FD BS serves multiple HD users simultaneously. However, since FD systems are plagued by severe self-interference (SI), we focus on the design of robust transceivers, which can cancel the residual SI left after antenna and analog cancellations. In particular, we address the sum mean-squared-errors (MSE) minimization problem by transforming it into an equivalent semidefinite programming (SDP) problem. We propose iterative alternating algorithms to design the transceiver matrices jointly and accordingly show the gains of FD over HD systems. We show that with proper SI cancellation, it is possible to achieve gains on sum rate of up to 70-80% over HD systems.