Industrial standards such as IS-856 [4] and 3GPP R5 [5] are multiuser scheduling systems supporting high speed downlink packet data access. In this dissertation, we adopt IS-856 as a reference model to investigate the characteristics of the multiuser scheduling system.
user 1
User 2
scheduled data in time slot
user K
User 5 User 3
User 8 User 1
: downlink signal : uplink feedback
Figure 2.5: A downlink multiuser scheduling system with TDMA protocol.
2.2.1 Scheduling Technique and Multiuser Diversity
Consider a multiuser scheduling system with the base station serving K downlink users as shown in Fig. 2.5. The basic operations of the multiuser scheduling system supporting high speed downlink packet data access are described as follows [48].
• The time-division-multiple-access (TDMA) protocol is adopted for the base station to service one user at a time slot.
• Each user measures and tracks its channel condition via the downlink common pilot signal and reports back to the base station through the uplink feedback channel.
• With the channel information from the feedback of all users, the base station determines to service one target user according to certain scheduling policies.
• Once the target user is selected, the base station uses the rate control adaptive modu-lation to transmit as many information bits as possible to the target user with its full transmit power.
By taking advantage of delay-tolerant data characteristics, the scheduling technique can extract the multiuser diversity from such the multiuser system to improve spectral efficiency [7]. The multiuser diversity gain can be explained as an analogy of the water-filling principle
across multiple users: pouring more resources to the user with better channel quality [8].
While allocating more resources to the user with better channel quality, the system may immediately face another key issue – how to schedule the transmissions for other users whose channel qualities are poor. Therefore, one of the major challenges in designing wireless scheduling algorithms is to achieve higher total system throughput but without sacrificing fairness to individual users too much.
A literature survey regarding the wireless scheduling technique is discussed as follows.
[49, 50] modified the fair scheduling policies used in traditional wireline systems to the wireless. However, their scheduling algorithms assume a simple two-state on-off channel model, which may not be able to fully capture all the wireless channel characteristics. In [9, 10, 51], several wireless scheduling algorithms were evaluated by Monte Carlo simulations for practical radio channels. In [52] and [53], it was shown that one-by-one time division scheduling scheme is better than the code division scheme from the standpoints of higher energy efficiency and better received signal quality, respectively.
2.2.2 Scheduling for Multiuser MIMO Systems
The multiuser scheduling system in Fig. 2.5 can be extended to the multiuser MIMO system.
A multiuser MIMO system consists of the base station employed with Nt transmit antennas and K multiple users employed with Nr receive antennas each. Accordingly, each link between the base station and individual user constitutes an (Nt, Nr) MIMO system.
Generally speaking, scheduling is a media access control (MAC) layer technique to deliver multiuser diversity by exploiting independent channel variations among user population.
By contrast, the multiple antenna technique is a physical layer approach to improve the performance of wireless links. A probe into the cross-layer interaction between the multiple antenna technique and multiuser scheduling has recently attracted attentions in the research community. In particular, a literature survey associated with the topic of scheduling for multiuser MIMO systems is discussed as follows.
• A number of works that used the fading mitigation based antenna techniques for the multiuser scheduling system can be found, for example, in [11, 51] and [54]-[58]. In [11], Viswanath and Tse proposed an opportunistic transmission scheme to increase the capacity of the multiuser scheduling system requiring only limited feedback. In [51], the authors conducted computer simulations to evaluate the capacity of the multiuser MIMO system with the ST and STBC antenna schemes. In [54], a semi-analytical result of spectral efficiency with the MRC antenna scheme and K-order multiuser diversity was derived. In [55], the impact of multiuser scheduling on the STBC methods was discussed in terms of the receive SNR distribution. Also, information theoretical treatments for the multiuser MIMO system with antenna diversity were provided in [56, 57, 58].
• Relatively, fewer works have taken advantage of scheduling techniques to improve the performance of the multiuser MIMO system with multiplexing-based antenna schemes, such as [59, 60]. Through simulation, the authors in [59] showed that multiuser di-versity can provide additional capacity gain for the multiplexing-based MIMO system with linear receivers. In [60], the authors harnessed multiuser diversity to enhance the capacity of the multiplexing-based MIMO systems using a random beamforming technique.
• When the base station is allowed to simultaneously transmit multiple beams to dif-ferent users (that is, not restricted to the TDMA protocol), both the temporal and spatial (antenna) domains can be exploited by scheduling to provide higher selection diversity order [61]-[64]. Because the signals to be transmitted for the multiple users are interfered with each other, the spatial-temporal domain scheduling is usually com-bined with additional pre-transmit signal processing techniques such as dirty-paper coding to achieve better performance [65]-[67].
In Chapter 3, we will introduce a unified analytical framework to investigate the capacity of the multiuser MIMO system with various diversity-based antenna schemes. In Chapters
4 and 5, we will discuss the topic of combining the multiplexing-based antenna scheme with the multiuser scheduling system.