Electrical and Electronic Engineering - Theses

Permanent URI for this collection

Search Results

Now showing 1 - 1 of 1
  • Item
    Thumbnail Image
    Enhanced user scheduling in MU-MIMO broadcast channels
    Ramanathan, Nikeeth ( 2013)
    Multi-user Multiple-Input Multiple-Output (MU-MIMO) systems are becoming increasingly important in wireless telecommunication networks. The multiple antennas at the base station allow for multiple users to be scheduled at the same time. This leads to multiplicative gains in transmission rates when compared to point-to-point communication. However, in order to reap the full benefits of the MU-MIMO, accurate channel state information (CSI) is required. This is a bigger issue in downlink broadcast channels (BC) as attaining accurate CSI would require users to feedback large amounts of information leading to large bandwidth requirements. As a result, this would restrict the data transmission rate in the uplink. Therefore, it is important to focus on limited or finite-rate feedback (LF/FRF) schemes. In limited feedback, the CSI feedback is restricted by the number of feedback bits allocated by each user. By limiting the feedback bits, a more practical scenario could be modelled whereby users utilize low complexity codebooks to feedback information with minimal delay. The trade-off that comes with this is saturation in the performance of the MU-MIMO BC. As the transmit power or signal-to-noise ratio (SNR) is increased, inter-user interference becomes a major problem. While transmission schemes such as zero-forcing beamforming (ZFBF) aid to resolve the performance to an extent, the limited CSI at the transmitter (CSIT) would imply that this interference cannot be completely subdued. In order to overcome this performance saturation, in this thesis, we investigate multi-user transmission and user-scheduling schemes with limited feedback. Scheduling of users is a critical and challenging problem. Base stations can attempt to maximize the performance of the broadcast channel by choosing the best set of users. However, in order to do so, an exhaustive search over all possible user-sets is required. This is generally computationally infeasible. Therefore, sub-optimal algorithms are required. While classical algorithms like semi-orthogonal user scheduling (SUS) increase the performance of schemes utilizing ZFBF, the performance still saturates. This thesis develops novel user-scheduling schemes to overcome the performance limits experienced in MU-MIMO BC. Our investigations identify the problems with classical schemes in their attempt to schedule as many users as possible. However, in the case when the channel is interference limited it is often better to switch to a smaller user-set in order to maximize performance. This raises an important research question of "how many users should the base station select?". Some attempts to answer this question have been made recently by employing multi-mode user scheduling. Here, each user approximates the system rate for different settings in order to work out its preferred number of co-scheduled users (mode of operation). The base station receives the preferred mode of operation of all the users along with CSI via limited feedback and makes a decision on the number of users to schedule. The main contribution of this thesis is the derivation of novel closed form expressions for rate approximations employed to determine their preferred mode in two different schemes. The first scheme outperforms the existing multi-mode scheduling schemes by providing improved scheduling criteria. The second is a novel scheme that combines SUS with multi-mode scheduling. Experimental simulations show the significant performance gains achieved by these schemes and reveal important considerations for future research.