Generally speaking, the new lines of research from multiuser MIMO broadcast systems are classified into three categories:
• First, rather than using the optimal DPC, the suboptimal but more practical MIMO broadcast schemes were proposed [4,27,46], such as the ZF-DPC, ZF beamforming, and orthogonal random beamforming. Because the DPC MIMO broadcast system faces the serious computation complexity issue, it requires huge amount of feedback information.
These suboptimal schemes can asymptotically achieve the same throughput of DPC when the number of users approaches to the infinity.
• Second, another important research direction for MIMO broadcast systems is to investi-gate the impacts of limited CSI due to the finite-rate or erroneous CSI feedback [47–55].
In [48], it was shown that the feedback load per user must be scaled together with both the number of transmit antennas as well as the system SNR to achieve the full multi-plexing gain with the near-perfect CSI.
• Third, MIMO broadcast transmission strategies were also applied to the multi-cellular scenario to cancel the inter-cell interference for improving spectral efficiency [17–21].
For the first category, two types of suboptimal MIMO broadcast systems were proposed in the literature: (1) the orthogonal random beamforming [46, 56–58] and (2) ZF based
beamforming [4].
• Firstly, recent research works regarding the orthogonal random beamforming for MIMO broadcast systems are briefly introduced as follows. In [46], it was proved that the or-thogonal random beamforming strategy can asymptotically achieve the same through-put slope of DPC when the number of users increases. To solve the difficulty of calculating the random beamformer’s weights for a large number of users, some low-complexity random beamformer approaches were proposed in [56–58].
• Secondly, we introduce the recent research results about the ZF-based beamforming for MIMO broadcast systems. In [4], a QR-based ZF-DPC MIMO broadcast system was proposed to maximize the sum rate of the MIMO broadcast system. Furthermore, the channel-inverse-based ZF beamforming was also proposed in [4], which is easier in calculating the beamforming weights than the ZF-DPC scheme. However, the ef-fects of user selection and user ordering were not considered in [4], and the number of users is assumed to be smaller than that of transmit antennas. Thus, many re-searches aimed to relax this assumption and examine a more general MIMO broadcast system when the number of users is larger than the number of transmit antennas. The authors of [59] proposed a greedy user-selection procedure for the ZF-DPC MIMO broadcast systems. In [60], it was shown that the slope of throughput against SNR in dB for the greedy ZF-DPC MIMO broadcast system is the same as that for the capacity-achieving DPC strategy. In [28], it was proved that the channel-inverse-based transmit ZF beamforming combined with multiuser scheduling can asymptotically ap-proach the capacity of the DPC-type MIMO broadcast system when the number of users approaches infinity. To overcome the prohibitively high complexity of exhaus-tively searching users, [28, 59–62] proposed low-complexity and effective user selection approaches for the MIMO broadcast systems.
Generally speaking, the objective of orthogonal random beamforming is to select a group of users to maximize their SINR according to partial CSI, whereas ZF-based beamforming is to nullify the mutual interference among users according to complete CSI.
For multi-cellular systems, the conventional widely-used inter-cell interference mitigation techniques are frequency reuse and FFR. The former scheme avoids utilizing the same
fre-quencies in the neighboring cells, while the later scheme allows universal frequency reuse for cell-center users. Conventional frequency reuse scheme yields lower spectrum utilization due to fewer available channels in each cell. To reduce the impact of frequency reuse on the throughput for each base station, the FFR scheme (or called reuse partition) assigns a larger frequency reuse factor for the cell-edge users and a smaller frequency reuse factor for the cell-center users. For example, Fig. 2.1 shows a conventional FFR planning in a cellular system with omni-cells [63]. The whole frequency band is partitioned into different zones, including the inner cell region with inner frequency band fA and the outer cell region with outer frequency bands fB. The outer frequency bands fB is further partitioned into three subbands fB1, fB2, and fB3. Under this framework, signal quality at cell edge can be improved at the cost of lower spectrum usage. To maintain spectrum efficiency among each cell, FFR principle is applied to the case of a sectorized cellular system as shown in Fig.
2.2 [64]. Because the inter-cell interference problem in the orthogonal frequency division multiple access (OFDMA) multi-cellular systems is more serious than that in the code di-vision multiple access (CDMA) systems, FFR once again becomes an important option for the next generation OFDMA broadband cellular mobile systems [63].
Clearly, combining the network MIMO and FFR techniques can have the advantages of complexity reduction and throughput enhancement. The joint FFR-based network MIMO system can execute the multi-base-station joint transmission only for the cell-edge users, and can apply the simple distance separation method for the cell-center users. Thus, the FFR-based network MIMO system can provide the sufficient SINR performance and avoid executing the joint multi-BS transmission at all time. Additionally, the universal frequency reuse for the cell-center users can improve the throughput due to higher trunking efficiency of the assigned channels. That is, the geographical locations of cells as well as mobiles now become another degree of freedom to be exploited for improving the performance of wireless systems. However, the study of network MIMO techniques on top of the FFR cellular system is rarely seen in the literature. Although both FFR and multi-base-station joint transmission are already considered for possible inter-cell interference cancelation techniques in the 3GPP LTE-A and the IEEE 802.16m WiMAX standards, to our knowledge, how to effectively integrate network MIMO with FFR is still an open issue.
In general, MIMO-OFDM systems can be classified into two types.
0
Figure 2.1: An example of FFR in a cellular system with omni-cells.
B1
Figure 2.2: An example of FFR in a cellular system with sectorized cells.
• Diversity-based MIMO-OFDM systems: to improve link reliability by exploiting the spatial and frequency diversity gains without CSI at the transmitter [40–45].
• Multiplexing-based MIMO-OFDM systems: to increase capacity by exploiting multi-plexing gain in the spatial domain [38]. The capacity of OFDM-based multimulti-plexing- multiplexing-based MIMO systems was investigated in [38]. It was shown that MIMO frequency-selective fading channels can provide higher ergodic capacity and outage capacity than MIMO flat-fading channels when delay paths can increase the total angular spread. If perfect CSI is also available at both the transmitter and receiver, the singular value decomposition (SVD)-based MIMO-OFDM systems, can decouple the MIMO chan-nel matrix and reduce the subchanchan-nel interference with the pre-processing and post-processing filters at the transmitter and receiver, respectively [65–68].
Compared with the diversity-based MIMO-OFDM systems, the spatial multiplexing based MIMO-OFDM systems have higher capacity in the inner region of a cell, but face more severe link reliability and coverage issues in the edge of cell. This is because diversity is deficient in the spatial multiplexing based MIMO system and the transmit power is split over multiple transmit antennas. In general, it is desirable to design a high-capacity wireless system with large coverage. In the literature, most MIMO-OFDM papers focus more on the capacity issue [38], [65–70] than the link quality improvement and coverage issues.
To improve coverage performance of MIMO-OFDM systems, a few approaches can be considered. The simplest way is to increase the transmit power. This approach is not desirable in general. The other approach is to design more powerful space-time codes or space-frequency codes with higher coding gain [45]. These kind of approaches are not suitable for the spatial multiplexing based MIMO-OFDM systems. Thus, we turn to exploit the user domain by developing an effective multiuser scheduling algorithm to improve the coverage performance of the spatial multiplexing based MIMO-OFDM systems.
Scheduling can enhance the performances of multiuser MIMO systems from different aspects [6, 26, 30, 59]. First, multiuser scheduling was designed to raise the sum-rate capac-ity of multiuser MIMO broadcast systems [6, 59]. Secondly, [30] showed that a max-min multiuser scheduling can enhance coverage as well as capacity for the spatial multiplexing based MIMO systems. Furthermore, multiuser scheduling integrated with the simple
zero-forcing MIMO receiver can approach the capacity of the optimal MIMO receiver [24–26].
Resource allocation for multiuser MIMO-OFDM systems is also an important research area because multiuser diversity, frequency diversity, and spatial diversity can be exploited si-multaneously. Various scheduling algorithms for multiuser MIMO-OFDM systems were pro-posed for different objectives, including maximizing data rates [71–74] (for SISO-OFDM systems) [65–70] (for MIMO-OFDM systems) and minimizing transmit power [75–78]. How-ever, to our knowledge, few radio resource management studies address the coverage issue for the spatial multiplexing based multiuser MIMO-OFDM systems.