Multi-user wireless communication systems with rate constraints

Abstract

In this thesis, we will make a comprehensive study on multi-user wireless communication systems where every user in the system must transmit a certain amount of information within each frame. This hard-fairness scenario is applicable to delay-sensitive services such as voice and video. We will discuss both the theoretical and practical aspects of multi-user wireless systems under such hard rate constraints. The main contributions of this thesis are listed below. In the first contribution, the closed-form expressions are derived for the average unconstrained minimum transmitted sum power (MTSP) of multi-user singleinput single-output (SISO) systems over multiple access channels (MACs) and broadcast channels (BCs). It is shown that significant performance gain, which is referred to as multi-user gain (MUG) in this thesis, can be achieved by allowing multi-user concurrent transmission. The MUG in SISO systems mainly comes from the near-far diversity among users and a large portion of MUG can be achieved with a small number of simultaneous users. The practical implementation aspect of multi-user SISO systems is also considered. We adopt interleave-division multiple-access (IDMA) as a platform and propose several power allocation methods to enhance the system performance. Both evolution and simulation results show that considerable MUG that is predicted in theory can indeed be achieved in practical environments. In the second contribution, we extend the results for the SISO scenario to the multiple-input multiple-output (MIMO) scenario. We avoid the complicated computation in finding the exact unconstrained MTSP of multi-user MIMO systems for each channel realization by adopting bounding techniques and derive the closed-form expression for the corresponding asymptotical average unconstrained MTSPs. We point out that, besides the near-far diversity, MUG in MIMO systems also comes from direction diversity that is provided by multiple antennas at the base station (BS), and the number of antennas at the BS has a more significant effect on the system capacity than that at the user side. In the meanwhile, we propose a low-cost but asymptotically optimal technique, i.e., the maximum eigenmode beamforming (MEB) technique, to realize the aforementioned MUG in multi-user MIMO systems with practical coding. In the third contribution, we consider the capacity analysis of cellular systems with various BS cooperation strategies. Based on the results in the first two contributions, some lower and upper bounds are derived for cellular systems with full BS cooperation (FBSC) and partial signal utilization (PSU). We show that, similar to the single-cell case considered in the second contribution, the number of antennas at each BS still has a more significant effect on the capacity of cellular systems. In the final contribution, we analyze the performance of multiple access systems with equal power allocation (EPA), i.e., controlling the received signal power of all users to the same level. The EPA scheme has much lower complexity than the general unequal power allocation (UPA) ones that have computational cost increasing rapidly with the number of users and must be implemented online for each channel realization. We study the feasibility and optimality of EPA and the corresponding system throughput (note that the throughput of a multiple access system with EPA is interference limited). We show that, although EPA is indeed sub-optimal in ideally coded systems, it can be optimal when practical coding is concerned, and the corresponding system throughput can be increased at the cost of some degree of distortion (e.g., a non-zero bit-error rate (BER) or frame-error rate (FER)). As for MIMO systems, we show that, provided that the MEB approach with EPA (i.e., MEB-EPA) is sufficiently good to achieve near-optimal performance when the number of users in concurrent transmission is large and the system sum rate is less than a certain threshold that increases approximately linearly with the number of receive antennas. Hence the throughput supportable by MEB-EPA can be very high provided that the receive antenna number is sufficiently large. Additionally, we also study the impact of imperfect channel state information (CSI) at the transmitter side (CSIT) on the performance of systems with EPA. We show that the EPA scheme is more robust to CSIT error than other alternatives such as TDMA, i.e., the former requires less additional power than the latter in combating the CSIT error to guarantee the same system performance as that without CSIT error. In summary, this thesis presents a comprehensive study on multi-user wireless communication systems with rate constraints. Both analytical and simulation results show that non-orthogonal CDMA-type multi-user concurrent transmission (e.g., IDMA) is advantageous over the orthogonal ones (such as TDMA) in terms of MUG, complexity and robustness.