## A Comparison of Soft-In/Soft-Out Algorithms for Turbo-Detection (1994)

Venue: | in Proc. Intern. Conf. on Telecomm |

Citations: | 35 - 4 self |

### BibTeX

@INPROCEEDINGS{Bauch94acomparison,

author = {Gerhard Bauch and Volker Franz},

title = {A Comparison of Soft-In/Soft-Out Algorithms for Turbo-Detection},

booktitle = {in Proc. Intern. Conf. on Telecomm},

year = {1994},

pages = {259--263}

}

### Years of Citing Articles

### OpenURL

### Abstract

In turbo--detection the "turbo principle" is applied to joint equalization and decoding. The performance of a turbo scheme strongly depends on the quality of the soft values passed between the soft--in/soft-- out decoders. In this paper we describe the differences between optimum and suboptimum soft--in/soft--out algorithms for equalization and decoding and compare them in a turbo--detection scheme concerning complexity and performance for perfect and mismatched channel estimation. Furthermore, a possibility to improve the soft--values of suboptimum algorithms which tend to be too optimistic is mentioned. I. INTRODUCTION AND PRINCIPLE OF TURBO--DETECTION The "turbo principle" which was first applied to parallel concatenated convolutional codes ("turbo--codes") in [1] can be applied to many detection and decoding problems. The idea of turbo--codes is to build a strong code by concatenation of simple component codes so that decoding can be performed in steps using algorithms of managea...

### Citations

1286 |
Optimal decoding of linear codes for minimizing symbol error rate
- Bahl, Cocke, et al.
- 1974
(Show Context)
Citation Context ...e optimum algorithm concerning minimization of bit error rate and computation of a posteriori information is the BCJR 1 --"Symbol by Symbol"--MAP algorithm (BCJR-- 1 Bahl, Cocke, Jelinek, Ra=-=viv MAP) ([4]-=-,[5],[3]). However, the complexity of the BCJR-- MAP is very high. Therefore, suboptimal solutions have to be considered, e.g. the BCJR--Max--Log--MAP ([6], [7]), the Soft--Output--Viterbi--Algorithm ... |

491 | Iterative decoding of binary block and convolutional codes
- Hagenauer, Offer, et al.
- 1996
(Show Context)
Citation Context ...timum algorithm concerning minimization of bit error rate and computation of a posteriori information is the BCJR 1 --"Symbol by Symbol"--MAP algorithm (BCJR-- 1 Bahl, Cocke, Jelinek, Raviv =-=MAP) ([4],[5]-=-,[3]). However, the complexity of the BCJR-- MAP is very high. Therefore, suboptimal solutions have to be considered, e.g. the BCJR--Max--Log--MAP ([6], [7]), the Soft--Output--Viterbi--Algorithm (SOV... |

272 |
A comparison of optimal and sub-optimal MAP decoding algorithms operating in the log domain
- Robertson, Villebrun, et al.
- 1995
(Show Context)
Citation Context ...R-- 1 Bahl, Cocke, Jelinek, Raviv MAP) ([4],[5],[3]). However, the complexity of the BCJR-- MAP is very high. Therefore, suboptimal solutions have to be considered, e.g. the BCJR--Max--Log--MAP ([6], =-=[7]-=-), the Soft--Output--Viterbi--Algorithm (SOVA) ([8]), and the Soft--Output--Viterbi--Equalizer (SOVE) ([6],[9]). In this paper we will describe the differences between these algorithms and compare the... |

258 |
A viterbi algorithm with soft-decision outputs and its applications
- Hagenauer, Hoher
- 1989
(Show Context)
Citation Context ...). However, the complexity of the BCJR-- MAP is very high. Therefore, suboptimal solutions have to be considered, e.g. the BCJR--Max--Log--MAP ([6], [7]), the Soft--Output--Viterbi--Algorithm (SOVA) (=-=[8]-=-), and the Soft--Output--Viterbi--Equalizer (SOVE) ([6],[9]). In this paper we will describe the differences between these algorithms and compare them in a turbo--detection--scheme concerning performa... |

232 |
Iterative correction of intersymbol interference: Turbo equalization
- Douillard, Jezequel, et al.
- 1995
(Show Context)
Citation Context ...processed output of the outer or parallel decoder is used as a priori input for the next iteration - similar to a turbo engine - this feedback is called the "turbo--component". In "turb=-=o--detection" ([2]-=-, [3]) the mobile radio channel with intersymbol--interference (ISI) including transmit and receive filters is regarded as a (time--varying) code, serially concatenated to the channel encoder which in... |

128 | The Turbo Principle: Tutorial Introduction and State of the Art - Hagenauer - 1997 |

38 | Iterative equalization and decoding in mobile communications systemsâ€ť, 2nd European Personal Mobile Communications Conference, Sept./Oct
- Bauch, Khorram, et al.
- 1997
(Show Context)
Citation Context ...ssed output of the outer or parallel decoder is used as a priori input for the next iteration - similar to a turbo engine - this feedback is called the "turbo--component". In "turbo--de=-=tection" ([2], [3]-=-) the mobile radio channel with intersymbol--interference (ISI) including transmit and receive filters is regarded as a (time--varying) code, serially concatenated to the channel encoder which in mobi... |

37 |
Optimum and sub-optimum detection of coded data disturbed by time-varying inter-symbol interference
- Koch, Baier
- 1990
(Show Context)
Citation Context ... (BCJR-- 1 Bahl, Cocke, Jelinek, Raviv MAP) ([4],[5],[3]). However, the complexity of the BCJR-- MAP is very high. Therefore, suboptimal solutions have to be considered, e.g. the BCJR--Max--Log--MAP (=-=[6]-=-, [7]), the Soft--Output--Viterbi--Algorithm (SOVA) ([8]), and the Soft--Output--Viterbi--Equalizer (SOVE) ([6],[9]). In this paper we will describe the differences between these algorithms and compar... |

25 |
Improved decoding with the SOVA in a parallel concatenated (turbo code) scheme
- Papke, Robertson
- 1996
(Show Context)
Citation Context ...th expanded memory due to the better quality of the L--values the performance of turbo--detection can be better than when using a SOVA equalizer despite the BER after the equalizer is still worse. In =-=[11]-=- for the AWGN--channel it was proposed to improve the performance of turbo--decoding with the SOVA by multiplying the extrinsic information by the factor c = m e 2 oe 2 e , where m e and oe 2 e are th... |

12 |
Near shannon limit error-correcting and decoding
- Berrou, Glavieux, et al.
- 1993
(Show Context)
Citation Context ...end to be too optimistic is mentioned. I. INTRODUCTION AND PRINCIPLE OF TURBO--DETECTION The "turbo principle" which was first applied to parallel concatenated convolutional codes ("tur=-=bo--codes") in [1]-=- can be applied to many detection and decoding problems. The idea of turbo--codes is to build a strong code by concatenation of simple component codes so that decoding can be performed in steps using ... |

9 | Iterative Equalization and Decoding for the GSM-System
- Bauch, Franz
- 1998
(Show Context)
Citation Context ...iterations 1 iteration 2 iterations 3 iterations Fig. 5: BER in class 1 for turbo--detection with perfect and mismatched channel estimation, BCJR--Max--Log--MAP-- equalizer and decoder time slot (see =-=[12]-=-). We transmitted data over a time invariant channel with taps g 0 = 0:227; g 1 = 0:46; g 2 = 0:688; g 3 = 0:46; g 4 = 0:227. Results for turbo detection with perfect channel knowledge are compared to... |

1 |
A soft--output Viterbi equalizer employing expanded memory length in a trellis
- Nagayasu, Kubo, et al.
- 1997
(Show Context)
Citation Context ...Therefore, suboptimal solutions have to be considered, e.g. the BCJR--Max--Log--MAP ([6], [7]), the Soft--Output--Viterbi--Algorithm (SOVA) ([8]), and the Soft--Output--Viterbi--Equalizer (SOVE) ([6],=-=[9]-=-). In this paper we will describe the differences between these algorithms and compare them in a turbo--detection--scheme concerning performance in bit error rate for perfect and mismatched channel es... |

1 |
Iterative channel estimation for "turbo"--detection
- Franz, Bauch
- 1998
(Show Context)
Citation Context ...he gain achieved by turbo--detection is reduced when channel estimation is considered. However, we can achieve an additional gain when channel estimation is improved during the iterations as shown in =-=[13]-=-. Figure 6 shows similar results for turbo--detection with a SOVE--equalizer and a SOVA-- 1.0e-04 1.0e-03 1.0e-02 1.0e-01 1.0e+00 2 4 6 8 10 12 Es/No in dB 2.3 dB 2.7 dB 2.9 dB estimated: 0 iterations... |