Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.486399
Title: Robust wireless MIMO transmission of video using multiple descriptions
Author: Tesanovic, Milos
ISNI:       0000 0001 3516 3481
Awarding Body: University of Bristol
Current Institution: University of Bristol
Date of Award: 2008
Availability of Full Text:
Access from EThOS:
Abstract:
Recent research has devised a way of increasing the achievable data rates for wireless communication by deploying multiple antennas at the transmitter and receiver, thereby creating multiple paths between the two ends of the system. This technique is known as MIMO (multiple-input-multiple-output) and its inclusion into the forthcoming wireless standards is already under way. Because of the promised increase in spectral efficiency, MIMO technology holds great appeal for bandwidth-hungry multimedia transmission. Conventional, single-description coding schemes (SDC), coupled with generic data transmission techniques (packet demultiplexing onto the created sub-channels) are shown in this thesis to yield unsatisfactory results, failing to harness the potential benefits arising from MIMO. As a way of remedying the underlying mapping problem, this thesis proposes the use of multiple-description coding (MDC) as a suitable video decomposition, with compelling results achieved for a wide range ofMIMO techniques and channel conditions. In the first instance, video transmission over MIMO systems that employ singular value decomposition (SVD) is studied and comparisons between SDC and proposed MDC are performed. MIMO-SVD systems are straightforward to simulate and provide greater mapping flexibility. The promising performance of MDC is then further enhanced by investigating power allocation in MIMO-SVD systems. Inspired by the water-filling power allocation strategy, results demonstrate that further improvements over SDC are possible if, for low SNRs, the stronger sub-channel is boosted at the expense of the weaker channel. For high SNR values, the converse is true. Using the underpinning multiple channels to send independent streams of data is known as spatial multiplexing (SM) and SVD is one way of achieving this. Various other SM techniques that do not require prior channel knowledge-CSI-at the transmitter exist and they are studied and compared with SVD in the context of video quality. The spatial correlation is changed by varying the angular width at the transmitter and receiver, with a view to emulating a wide range of possible deployment scenarios. Although not optimal for the entire range of SNR values studied, SVD dominates other SM techniques under study and it is therefore of great interest to investigate the effects of CSI latency in SVD systems. These are quantified and the severe degradation when CSI latency cannot be ignored is tackled. Analytical results justify the deployment of SVD even when prior CSI is not perfect, as it facilitates post-processing that is shown to restore results similar to those obtained when prior CSI is perfect. No study of MIMO systems would be complete without investigating possible improvements from space-time (ST) techniques. Even though SM seems an obvious choice for video transmission as it offers the much-needed increase in throughput, it is shown in this thesis that simple MIMO-STBC systems, while not offering any increase in spectral efficiency compared to SISO systems, can outperform SM systems in the presence of errors because of their superior error performance at low SNR. When SDC is used to transmit video over SM, the improvements from multiplexing are only visible in the quasi error-free range of SNR values. MDC is demonstrated to move the SMlSTBC cross-over point towards the region of lower SNR values, reducing the penalty that has to be paid for the higher throughput offered by SM systems. MDC is shown to equally be able to improve the performance of STBC systems by interleaving the descriptions prior to transmission. Further improvements arising from the use of cross-packet FEC are achieved. However, interleaving incurs delays that may not be acceptable. These delays are therefore quantified for different interleaving scenarios. It is shown that MDC can help reduce the decoding delays, while at the same time offering improvements not possible from SDC interleaving, with or without FEC.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.486399  DOI: Not available
Share: