Loading…

An implementation friendly low complexity multiplierless LLR generator for soft MIMO sphere decoders

When combined with advanced FEC techniques such as the turbo code and LDPC code, soft-output MIMO sphere decoders significantly outperform hard-output sphere decoders. Hence, algorithms and implementations of soft-output sphere decoders have attracted intensive interest in recent years. Practical so...

Full description

Saved in:
Bibliographic Details
Main Authors: Min Li, Novo, D., Bougard, B., Naessens, F., Van Der Perre, L., Catthoor, F.
Format: Conference Proceeding
Language:English
Subjects:
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:When combined with advanced FEC techniques such as the turbo code and LDPC code, soft-output MIMO sphere decoders significantly outperform hard-output sphere decoders. Hence, algorithms and implementations of soft-output sphere decoders have attracted intensive interest in recent years. Practical soft-output sphere decoder implementations often consist of a list generator and a LLR generator. Most existing implementations focus on the list generator, and the LLR generator is implemented in a relatively straightforward way. However, the LLR generator accounts for a great part of the complexity. Our contribution is an implementation friendly low complexity multiplierless LLR generator. We apply selective and incremental updating, algebraic simplifications and strength reductions to reduce the algorithmic complexity and to eliminate all multiplications. When integrated with the SSFE list generator, our scheme not only remove 100% multiplications, but also remove 26% to 83% additions, 76% to 94% bit-shifts and 63% to 91% memory operations. Besides the algorithmic aspects, we extract the key data-flow block with well-defined control signals. This can be easily mapped onto micro-architectures and implemented as the data-path in ASICs, or a function unit in ASIPs.
ISSN:2162-3562
2162-3570
DOI:10.1109/SIPS.2008.4671748