WPR.4 - Iterative Receivers for Wireless Communications

Objectives

This workpackage seeks to develop and extend the theoretical framework for iterative processing, and devise methods for code design and optimized receiver design, thereby enhancing overall radio performance. The aim is to advance the theoretical foundations of iterative processing, while tackling all aspects of practical design that accompany an implementation of iterative systems, from imperfect channel knowledge to fixed point implementations and down to synchronization problems and joint receiver design.

Description of work

Research in this workpackage is organized along 3 tasks:

Task TR4.1 Polar Coding

This task investigates the practical uses of a new coding method known as “polar coding”, developed by WPR4 participants. This method is remarkable for providing one of the first constructive approaches to code design to approach the capacity of specific channels. However, while this constitutes a notable theoretical breakthrough, in order to be of practical relevance a number of extensions to the known theory are necessary, which are being investigated within WPR4: a) low complexity decoding methods; b) extensions to more general channel models; c) codes over non-binary alphabets, including real and complex numbers; d) applications of polar codes beyond error control coding.

Main Players: Bilkent, CNIT, CNRS

Task TR4.2 Joint Receivers

This task concentrates on the design of receivers where iterative processing is applied throughout, beyond its traditional use in turbo and Low-Density Parity-Check (LDPC) decoding. This includes in particular: a) drawing parallels between iterative divergence minimization based approaches used in channel estimation and detection and iterative decoding techniques, and applying the resulting principles to joint iterative receivers; b) applying iterative techniques to synchronization and phase noise estimation and including them in a joint iterative receivers; c) applying iterative techniques to multi-user scenarios; d) develop analysis methods based on information geometry and applying them to the analysis of joint receivers.

Main players: AAU, UCL, RWTH, FTW, CNRS, UoC

Task TR4.3 Reduced Complexity Decoding

This task concentrates on the simplification of iterative techniques in view of making them implementable in practical radio receivers. Iterative decoding has a constant complexity per code symbol, or linear in the code block length. This constitutes a major breakthrough with respect to prior art where the complexity remained exponential in either the block length or the number of states in the encoder. Nonetheless, while complexity measures remain a central tool to evaluate the potential practicality of decoding methods, they only record the actual implementation complexity to within any multiple. In other words, they provide an important asymptotic measure of complexity for block lengths tending towards infinity, but for practical finite block length and real-life decoders they can be misleading. Therefore, there is ample room for improvement and fine-tuning of iterative decoding methods to reduce the complexity in practice and minimize the loss of performance. This task concentrates on the design of reduced complexity decoders for practical implementation, in particular: a) design of improved Log-MAP decoders for turbo codes; b) design of decoders for turbo trellis-coded modulation and and for turbo codes in conjunction with space-time coding; c) design of decoders operating in the fixed-point domain, including quantization effects and clipping; d) analysis and design of reduced complexity decoders based on the principles of mismatched decoding.

Main players: PUT, IASA, CNIT, UCam, CNRS, FTW, UoC

In addition to these three tasks, a number of transversal activities are pursued across tasks, notably the investigation of theory underlying iterative processing, for performance evaluation and prediction.

The WP structure has undergone a major re-organization at TO+18. The original WP description in the NEWCOM++ proposal contained 8 tasks organized along 4 major axes:

• The first axis targeted the theory of iterative processing in two tasks dedicated to analytical bounds and information combining. These activities are continuing as transversal activities across tasks as described in the previous paragraph.
• The second axis targeted new codes and decoders in two tasks dedicated to new codes and new decoding algorithms. The activities concerning new codes have been concentrated around polar codes. The activities concerning new decoding algorithms have been shifted from pure linear programming-based decoding to convex oprimization for joint estimation, detection and decoding and are continued in Task 1.
• The third axis targeted implementation issues in two tasks dedicated to mismatched decoding and fixed point decoding. This axis has been consolidated and re-structured into a single Task 3 in the current WP description.
• The fourth axis targeted iterative processing beyond decoding within two tasks dedicated to turbo synchronization and joint design and optimization. This axis has been consolidated and restructured into a single Task 2 in the current WP description.