Scarce system resources in wireless multiuser systems call for efficient use of power and bandwidth. This problem has several dimensions. For each communication link within a multiuser system, the problem is to design efficient, low-complexity source-channel coding algorithms -- for applications with delay constraints, this calls for joint source-channel coding, due to necessarily non-ideal coding algorithms. From a system perspective, the question is how to use the given resources (bandwidth, power) efficiently but still in such as way that the quality-of-service for all users is maximised. The conventional approach will ignore both channel knowledge as well as information about the applications in the medium access layer which is rather wasteful. While using channel knowledge when scheduling users for channel access will allow to achieve large multiuser diversity (transmit power) gains, knowledge about the applications will be crucial to still meet quality-of-service requirements. Joint Source-Channel Coding and Decoding: - Separation theorem of information theory - Practical limitations in real-world systems - Cooperation between source and channel decoding: bad-frame handling in speech transmission - Joint Source-Channel Coding Theory: the optimal system - Channel-Optimised Vector Quantisation - Optimised Index Mappings for Quantisers - Multiple Descriptions - Soft-In/Soft-Out Channel Decoding - Soft-Input Source Decoding - Iterative Source-Channel Decoding Cross-Layer Design: - Open Systems Interconnection Reference Model: basic concept and limitations for wireless systems - Cross-Layer Design: principles - Wireless multiuser systems: Information theoretical aspects - Multiuser diversity gain - Quality-of-Service vs. Throughput - Optimal resource allocation and user scheduling - Practical aspects, delay in particular - Scheduling for wireless: Round Robin, Proportional Fair, Content- and Channel-Aware Scheduling
Institut 389, Room: CG0118, Time: Tuesday, 16:00 - 18:00
First class: 09.10.2012
Ein Skriptum zur Lehrveranstaltung ist erhältlich: Details werden in der ersten Vorlesung bekanntgegeben. Course Text (recommended, not compulsory): Norbert Goertz, "JOINT SOURCE-CHANNEL CODING OF DISCRETE-TIME SIGNALS WITH CONTINUOUS AMPLITUDES", World Scientific / Imperial College Press, Sept. 2007 ISBN 978-1-86094-845-9
