A Low-complexity Near-ML Decoding Via Reduced Dimension Maximum Likelihood Search
Uloženo v:
| Název: | A Low-complexity Near-ML Decoding Via Reduced Dimension Maximum Likelihood Search |
|---|---|
| Autoři: | Jun Won Choi, Byonghyo Shim, Andrew C. Singer |
| Přispěvatelé: | The Pennsylvania State University CiteSeerX Archives |
| Zdroj: | http://ipl.korea.ac.kr/paper/rd_lts.pdf. |
| Rok vydání: | 2008 |
| Sbírka: | CiteSeerX |
| Témata: | Index Terms Maximum likelihood (ML) decoding, Sphere decoding, Minimum mean square error (MMSE, Multiple input multiple output (MIMO, Stack algorithm, Dimension reduction, List tree search |
| Popis: | In this paper, we consider a low-complexity ML detection technique referred to as reduced dimension ML search (RD-MLS). RD-MLS directly refers the division of symbols into two groups viz. strong and weak group for searching over vector space with strong symbols instead of whole symbols. The inevitable performance loss, due to the exclusion of weak symbols, is compensated by 1) the list tree search which is an extended version of single best searching algorithm called sphere decoding and 2) re-computation of weak symbols corresponding to each strong symbols found at the list tree search. Furthermore, in order to lessen the computational burden in list tree search, we employ tree pruning strategy that removes the unpromising subtrees before they are being searched. Two pruning techniques, called list sphere decoding with probabilistic pruning (LSD-PP) and list stack algorithm with probabilistic pruning (LSA-PP), is proposed for this purpose. From the simulations study on the multi-input-multi-output (MIMO) systems, we demonstrate that the RD-MLS shows near constant complexity over wide SNR range of interest (10−1 ∼ 10−4) while limiting the performance loss within a dB from the ML detection. |
| Druh dokumentu: | text |
| Popis souboru: | application/pdf |
| Jazyk: | English |
| Relation: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.506.7780; http://ipl.korea.ac.kr/paper/rd_lts.pdf |
| Dostupnost: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.506.7780 http://ipl.korea.ac.kr/paper/rd_lts.pdf |
| Rights: | Metadata may be used without restrictions as long as the oai identifier remains attached to it. |
| Přístupové číslo: | edsbas.61BCA1AB |
| Databáze: | BASE |
Nájsť tento článok vo Web of Science | Abstrakt: | In this paper, we consider a low-complexity ML detection technique referred to as reduced dimension ML search (RD-MLS). RD-MLS directly refers the division of symbols into two groups viz. strong and weak group for searching over vector space with strong symbols instead of whole symbols. The inevitable performance loss, due to the exclusion of weak symbols, is compensated by 1) the list tree search which is an extended version of single best searching algorithm called sphere decoding and 2) re-computation of weak symbols corresponding to each strong symbols found at the list tree search. Furthermore, in order to lessen the computational burden in list tree search, we employ tree pruning strategy that removes the unpromising subtrees before they are being searched. Two pruning techniques, called list sphere decoding with probabilistic pruning (LSD-PP) and list stack algorithm with probabilistic pruning (LSA-PP), is proposed for this purpose. From the simulations study on the multi-input-multi-output (MIMO) systems, we demonstrate that the RD-MLS shows near constant complexity over wide SNR range of interest (10−1 ∼ 10−4) while limiting the performance loss within a dB from the ML detection. |
|---|