Sitemap  |  Contact  |  Home  |  中文  |  CAS  |  Director’s Email
International Cooperation
Education & Training
Societies & Publications
Chinese Journal of Acoustics
 
 
  Location:Home>Research>Research Progress
An Approximate Message Passing Based Near-Capacity Method over Severe Channels
Author:
ArticleSource:
Update time: 2019/04/30
Viewed:
Close
Text Size: A A A
Print

 

For communication systems, the propagation of signals from the transmitter to the receiver via multiple paths is called multipath effect, leading to the overlap of multiple sequentially-transmitted symbols at the receiver. The phenomenon causes inter-symbol interference (ISI) at the receiver which makes it difficult for communication signal detection and symbol recovery.

In order to combat the ISI and approach the channel capacity, researchers from the Institute of Acoustics (IOA) of the Chinese Academy of Sciences along with Prof. TAO Jun from Southeast University proposed an approximate message passing (AMP)-based near-capacity method which significantly narrows the performance gap from the severe channel capacity.

The study has been published in the IEEE Access.

Turbo equalization is a well-known receiver technique for combating the ISI. It generally consists of a soft equalizer and a soft decoder which iteratively exchange extrinsic information to improve the detection performance. Typical turbo equalization schemes generally adopt a fixed convolutional code, a turbo code or a low density parity check code, which does not properly match the channel, as the channel code, and choose the suboptimal minimum mean square error (MMSE)-optimized linear equalizers and decision feedback-type equalizers as the equalization scheme. In summary, this turbo receiver has a large gap to the channel capacity.

Aiming at the soft equalizer, researchers proposed a semi-adaptive damping (SAD)-aided vector AMP (VAMP) soft frequency-domain equalizer (SFDE). It completely eliminates the ISI based on some a priori information over the low signal-to-noise ratio (SNR) region employing the fast Fourier transform with the quasi-linear computational complexity. Afterwards, researchers designed and further optimized the inner and outer codes based on the extrinsic information transfer (EXIT) curve of the SAD-VAMP-SFDE.

The inner code introduced the unity-rate code (URC) to eliminate the error floor of iterative decoding. The outer code was designed as an irregular convolutional code (IrCC) which matched the EXIT curve of the concatenated URC decoder and the SAD-VAMP-SFDE via a weight combination of multiple subcodes. It effectively lowered the iterative SNR-threshold. Finally, a three-fold iterative processing among the SAD-VAMP-SFDE, the URC decoder, and the IrCC decoder was achieved.

Simulation results showed that under the 6-path minimum-distance severe channel, the SNR gap of the proposed system with the SAD-VAMP-SFDE was only 1.3 dB from the approximate uniform input capacity.

 

Figure 1. Block diagram of proposed near-capacity transceiver system. (Image by IOA)

Figure 2. EXIT chart (left) and bit error rate performance curves (right) of the three-fold iterative receiver. (Image by IOA)

In the following study, researchers would apply this method to underwater acoustic communication (UAC) systems for improving its communication rate and performance as well as reducing power consumption.

Funding for this research came from the National Natural Science Foundation of China (No. 61471351, 61871114), and the National Key Research and Development Program of China (No. 2016YFC0300300).

Reference:

LI Dong, WU Yanbo, TAO Jun, ZHU Min. Performance Analysis and Improvement for VAMP Soft Frequency-Domain Equalizers. IEEE Access. 2019, 7: 42495-42506. DOI: 10.1109/ACCESS.2019.2907326.

Contact:

ZHOU Wenjia

Institute of Acoustics, Chinese Academy of Sciences, 100190 Beijing, China

E-mail: media@mail.ioa.ac.cn

 
Copyright ©2002-2008 Institute of Acoustics, Chinese Academy of Sciences
No. 21 North 4th Ring Road, Haidian District, 100190 Beijing, China
E-mail: ioa@mail.ioa.ac.cn