摘要
深海水声信道具有信噪比低、信道参数随时间和空间变化的特点,而且由于反馈延时长等限制,在当前的深海远程水声通信系统中很难建立有效的反馈信道,这也使基于反馈信道的差错控制方法和组网协议难以应用于深海远程水声通信.针对以上两方面问题,本文提出了一种基于Ludy Transform编码(LT编码)的联合迭代均衡译码方法,简称LT-Turbo均衡方法.该方法利用了LT编码的无固定编码速率、与信道和通信节点无关等特性,能够在深远海水声通信中灵活控制编码速率,并通过联合迭代均衡译码,进一步提高通信效率.数值仿真结果显示,相比于分离的均衡译码方法,LT-Turbo均衡方法能够获得更低的误码率,而且更接近理论极限;在低信噪比的深海水声信道中,实现同样的可靠信息传输,该方法所需的编码冗余度为传统时域重发方法的1/2左右,提高了有效信息传输速率.
Long range deep-water acoustic communication (LDAC) is challenging due to low SNR and time- space-varying channel parameters. Since there exists the long feedback delay, it's hard to construct the feedback channel in LDAC. So the feedback based error control methods and the link-layer framings are all disabled in LDAC. To solve these problems, the Ludy Transform code (LT code), which is a kind ofrateless fountain code, is introduced to replace the traditional channel code of the classical LDAC system. LT code brings in the rateless property and the channel independence which are the foundation of the no-feedback link-layer framing, but the performance of this receiver scheme has a distinct gap compared with the theoretical bound. So a LT-Turbo equalization method is proposed, where the adaptive linear equalization and the decoder of LT code are jointly optimized in an iterative process. The simulation results show that the LT-Turbo equalization could get a lower bit error rate than the classical method under various kinds of channels. The results of another contrastive simulation between the LT-Turbo equalization method and the classical time domain retransmission method show that the minimum required code redundancy of LT-Turbo equalization is only 1/2 of the classical time domain retransmission method in the same Iow-SNR deep-water acoustic channel, which proves the excellent performance of the LT-Turbo equalization.
出处
《中国科学:物理学、力学、天文学》
CSCD
北大核心
2016年第9期90-97,共8页
Scientia Sinica Physica,Mechanica & Astronomica
基金
国家自然科学基金资助项目(编号:11434012
41561144006
11404366)
关键词
深海
LT编码
水声通信
TURBO均衡
deep water, LT code, acoustic communication, Turbo equalization