In this paper, we describe a hard-decision decoding technique based on Genetic Algorithms (HDGA), which is applicable to the general case of error correcting codes where the only known structure is given by the genera...In this paper, we describe a hard-decision decoding technique based on Genetic Algorithms (HDGA), which is applicable to the general case of error correcting codes where the only known structure is given by the generating matrix G. Then we present a new soft-decision decoding based on HDGA and the Chase algorithm (SDGA). The performance of some binary and non-binary Linear Block Codes are given for HDGA and SDGA over Gaussian and Rayleigh channels. The performances show that the HDGA decoder has the same performances as the Berlekamp-Massey Algorithm (BMA) in various transmission channels. On the other hand, the performances of SDGA are equivalent to soft-decision decoding using Chase algorithm and BMA (Chase-BMA). The complexity of decoders proposed is also discussed and compared to those of other decoders.展开更多
This paper proposes a modification of the soft output Viterbi decoding algorithm (SOVA) which combines convolution code with Huffman coding. The idea is to extract the bit probability information from the Huffman codi...This paper proposes a modification of the soft output Viterbi decoding algorithm (SOVA) which combines convolution code with Huffman coding. The idea is to extract the bit probability information from the Huffman coding and use it to compute the a priori source information which can be used when the channel environment is bad. The suggested scheme does not require changes on the transmitter side. Compared with separate decoding systems, the gain in signal to noise ratio is about 0 5-1.0 dB with a limi...展开更多
Two modified BP algorithms related to vertical and horizontal processes are proposed to accelerate iterative low-density parity- check （LDPC） decoding over an additive white Gaussian noise （AWGN） channel, where th...Two modified BP algorithms related to vertical and horizontal processes are proposed to accelerate iterative low-density parity- check （LDPC） decoding over an additive white Gaussian noise （AWGN） channel, where the newly updated extrinsic information is immediately used in the current decoding round. Theoretical analysis and simulation results demonstrate that both the modified approaches provide significant performance improvements over the traditional BP algorithm with almost no additional decoding complexity. The proposed algorithm with modified horizontal process offers even better performance than another algorithm with the modified horizontal process. The two modified BP algorithms are very promising in practical communications since both can achieve an excellent trade-off between the performance and decoding complexity.展开更多
It is well known that turbo decoding always begins from the first component decoder and supposes that the apriori information is '0' at the first iterative decoding. To alternatively start decoding at two comp...It is well known that turbo decoding always begins from the first component decoder and supposes that the apriori information is '0' at the first iterative decoding. To alternatively start decoding at two component decoders, we can gain two soft output values for the received observation of an input bit. It is obvious that two soft output values comprise more sufficient extrinsic information than only one output value obtained in the conventional scheme since different start points of decoding result in different combinations of the a priori information and the input codewords with different symbol orders due to the permutation of an interleaver. Summarizing two soft output values for erery bit before making hard decisions, we can correct more errors due to their complement. Consequently, turbo codes can achieve better error correcting performance than before in this way. Simulation results show that the performance of turbo codes using the novel proposed decoding scheme can get a growing improvement with the increment of SNR in general compared to the conventional scheme. When the bit error probability is 10-5 , the proposed scheme can achieve 0.5 dB asymptotic coding gain or so under the given simulation conditions.展开更多
Iterative demodulation and decoding scheme is analyzed and modulation labeling is considered to be one of the crucial factors to this scheme. By analyzing the existent mapping design criterion, four aspects are found ...Iterative demodulation and decoding scheme is analyzed and modulation labeling is considered to be one of the crucial factors to this scheme. By analyzing the existent mapping design criterion, four aspects are found as the key techniques for choosing a label mapping. Based on this discovery, a novel mapping design criteflon is proposed and two label mappings are searched according to it. Simulation results show that the performance of BICM-ID using the novel mappings is better than the former ones. The extrinsic information transfer （EXIT） chart is introduced and it is used to evaluate the proposed mapping design criteria.展开更多
In this paper,we innovatively associate the mutual information with the frame error rate(FER)performance and propose novel quantized decoders for polar codes.Based on the optimal quantizer of binary-input discrete mem...In this paper,we innovatively associate the mutual information with the frame error rate(FER)performance and propose novel quantized decoders for polar codes.Based on the optimal quantizer of binary-input discrete memoryless channels(BDMCs),the proposed decoders quantize the virtual subchannels of polar codes to maximize mutual information(MMI)between source bits and quantized symbols.The nested structure of polar codes ensures that the MMI quantization can be implemented stage by stage.Simulation results show that the proposed MMI decoders with 4 quantization bits outperform the existing nonuniform quantized decoders that minimize mean-squared error(MMSE)with 4 quantization bits,and yield even better performance than uniform MMI quantized decoders with 5 quantization bits.Furthermore,the proposed 5-bit quantized MMI decoders approach the floating-point decoders with negligible performance loss.展开更多
Applications on iterative control and multiple input multiple output (MIMO) system were developed. Two new charts derived from extrinsic information transfer (EXIT) chart were employed as the designing tools, which ar...Applications on iterative control and multiple input multiple output (MIMO) system were developed. Two new charts derived from extrinsic information transfer (EXIT) chart were employed as the designing tools, which are called as output mutual information chart with defined iterative degree (DID) chart and near optimum output mutual information (NOMI) chart respectively. Different from the EXIT chart, they can show the iterative performance on the whole signal-to-noise ratio range with one single curve, whereas computation complexity is greatly reduced compared with conventional bit error ratio (BER) performance curve. The iterative control was implemented according to a near-optimum iterative degree vector determined by NOMI chart, the reasonability of uncertain parameters was analyzed in one MIMO system. The concepts were illustrated based on bit-interleaved coded modulation with iterative decoding (BICM-ID).展开更多
This paper proposes a novel mapping scheme for bit-interleaved coded modulation with iterative decoding(BICM-ID).The symbol mapping is composed of two QPSK with different radiuses and phases,called cross equalization-...This paper proposes a novel mapping scheme for bit-interleaved coded modulation with iterative decoding(BICM-ID).The symbol mapping is composed of two QPSK with different radiuses and phases,called cross equalization-8PSK-quasi-semi set partitioning(CE-8PSK-Quasi-SSP).Providing the same average power,the proposed scheme can increase the minimum squared Euclidean distance(MSED)and then improve the receiving performance of BICM-ID compared with conventional symbol mapping schemes.Simultaneously,a modified iteration decoding algorithm is proposed in this paper.In the process of iteration decoding,different proportion of the extrinsic information to the systematic observations results in distinct decoding performance.At high SNR(4~9dB),the observation information plays a more important role than the extrinsic information.Simulation results show that the proportion set at 1.2 is more suitable for the novel mapping in BICM-ID.When the BER is 10^(-4),more than 0.9dB coding gain over Rayleigh channels can be achieved for the improved mapping and decoding scheme.展开更多
Considering the influence of more random atmospheric turbulence, worse pointing errors and highly dynamic link on the transmission performance of mobile multiple-input multiple-output （MIMO） free space optics （FSO...Considering the influence of more random atmospheric turbulence, worse pointing errors and highly dynamic link on the transmission performance of mobile multiple-input multiple-output （MIMO） free space optics （FSO） communica- tion systems, this paper establishes a channel model for the mobile platform. Based on the combination of Alamouti space-time code and time hopping ultra-wide band （TH-UWB） communications, a novel repetition space-time coding （RSTC） method for mobile 2x2 free-space optical communications with pulse position modulation （PPM） is devel- oped. In particular, two decoding methods of equal gain combining （EGC） maximum likelihood detection （MLD） and correlation matrix detection （CMD） are derived. When a quasi-static fading and weak turbulence channel model are considered, simulation results show that whether the channel state information （CSI） is known or not, the coding sys- tem demonstrates more significant performance of the symbol error rate （SER） than the uncoding. In other words, transmitting diversity can be achieved while conveying the information only through the time delays of the modulated signals transmitted from different antennas. CMD has almost the same effect of signal combining with maximal ratio combining （MRC）. However, when the channel correlation increases, SER performance of the coding 2×2 system de- grades significantly.展开更多
文档分析与识别(简称文档识别)技术将各种非结构化文档数据(图像、联机笔迹)转化为结构化数据,便于计算机处理和理解,应用场景十分广阔。20世纪60年代以来,文档识别方法研究与应用受到广泛关注并取得巨大进展。得益于深度学习技术的发...文档分析与识别(简称文档识别)技术将各种非结构化文档数据(图像、联机笔迹)转化为结构化数据,便于计算机处理和理解,应用场景十分广阔。20世纪60年代以来,文档识别方法研究与应用受到广泛关注并取得巨大进展。得益于深度学习技术的发展和应用,文档识别的性能快速提升,相关技术在文档数字化、票据处理、笔迹录入、智能交通、文档检索与信息抽取等领域得到广泛应用。首先介绍文档识别的背景和技术范畴,回顾该领域发展历史,然后重点对深度学习方法兴起以来的研究进行综述,分析当前技术存在的不足,并建议未来值得重视的研究方向。研究现状综述部分,按文档分析与识别的几个主要技术环节(文档图像预处理、版面分析、场景文本检测、文本识别、结构化符号和图形识别、文档检索与信息抽取)分别进行介绍,简述传统方法研究的代表性工作,重点介绍深度学习方法研究的新进展。总体上,当前研究对象向深度、广度扩展,处理方法全面转向深度神经网络模型和深度学习方法,识别性能大幅提升且应用场景不断扩展。在现状分析基础上,指出当前技术在识别精度和可靠性、可解释性、学习能力和自适应性等方面还有明显不足。最后从提升性能、应用扩展、提升学习能力几个角度提出一些研究方向。从提升性能角度,研究问题包括文本识别可靠性、可解释性、全要素识别、长尾问题、多语言、复杂版面分割与理解、变形文档分析与识别等。应用扩展包括新应用(如机器人流程自动化(robotic process automation,RPA)、文字信息抄录、考古)和新技术问题(语义信息抽取、跨模态融合、面向应用的推理决策等)两方面。从提升学习能力角度,相关问题包括小样本学习、迁移学习、多任务学习、领域自适应、结构化预测、弱监督学习、自监督学习、开放集学习和跨模态学习等。展开更多
文摘In this paper, we describe a hard-decision decoding technique based on Genetic Algorithms (HDGA), which is applicable to the general case of error correcting codes where the only known structure is given by the generating matrix G. Then we present a new soft-decision decoding based on HDGA and the Chase algorithm (SDGA). The performance of some binary and non-binary Linear Block Codes are given for HDGA and SDGA over Gaussian and Rayleigh channels. The performances show that the HDGA decoder has the same performances as the Berlekamp-Massey Algorithm (BMA) in various transmission channels. On the other hand, the performances of SDGA are equivalent to soft-decision decoding using Chase algorithm and BMA (Chase-BMA). The complexity of decoders proposed is also discussed and compared to those of other decoders.
文摘This paper proposes a modification of the soft output Viterbi decoding algorithm (SOVA) which combines convolution code with Huffman coding. The idea is to extract the bit probability information from the Huffman coding and use it to compute the a priori source information which can be used when the channel environment is bad. The suggested scheme does not require changes on the transmitter side. Compared with separate decoding systems, the gain in signal to noise ratio is about 0 5-1.0 dB with a limi...
基金National Mobile Communication Research Laboratory,Southeast University(No.W200704),ChinaNatural Science foundation of Jiangsu Province (No.BK2006188),ChinaQuebec-China Joint Research Foundation by McGill University,Montreal,Quebec,Canada
文摘Two modified BP algorithms related to vertical and horizontal processes are proposed to accelerate iterative low-density parity- check （LDPC） decoding over an additive white Gaussian noise （AWGN） channel, where the newly updated extrinsic information is immediately used in the current decoding round. Theoretical analysis and simulation results demonstrate that both the modified approaches provide significant performance improvements over the traditional BP algorithm with almost no additional decoding complexity. The proposed algorithm with modified horizontal process offers even better performance than another algorithm with the modified horizontal process. The two modified BP algorithms are very promising in practical communications since both can achieve an excellent trade-off between the performance and decoding complexity.
文摘It is well known that turbo decoding always begins from the first component decoder and supposes that the apriori information is '0' at the first iterative decoding. To alternatively start decoding at two component decoders, we can gain two soft output values for the received observation of an input bit. It is obvious that two soft output values comprise more sufficient extrinsic information than only one output value obtained in the conventional scheme since different start points of decoding result in different combinations of the a priori information and the input codewords with different symbol orders due to the permutation of an interleaver. Summarizing two soft output values for erery bit before making hard decisions, we can correct more errors due to their complement. Consequently, turbo codes can achieve better error correcting performance than before in this way. Simulation results show that the performance of turbo codes using the novel proposed decoding scheme can get a growing improvement with the increment of SNR in general compared to the conventional scheme. When the bit error probability is 10-5 , the proposed scheme can achieve 0.5 dB asymptotic coding gain or so under the given simulation conditions.
文摘Iterative demodulation and decoding scheme is analyzed and modulation labeling is considered to be one of the crucial factors to this scheme. By analyzing the existent mapping design criterion, four aspects are found as the key techniques for choosing a label mapping. Based on this discovery, a novel mapping design criteflon is proposed and two label mappings are searched according to it. Simulation results show that the performance of BICM-ID using the novel mappings is better than the former ones. The extrinsic information transfer （EXIT） chart is introduced and it is used to evaluate the proposed mapping design criteria.
基金financially supported in part by National Key R&D Program of China(No.2018YFB1801402)in part by Huawei Technologies Co.,Ltd.
文摘In this paper,we innovatively associate the mutual information with the frame error rate(FER)performance and propose novel quantized decoders for polar codes.Based on the optimal quantizer of binary-input discrete memoryless channels(BDMCs),the proposed decoders quantize the virtual subchannels of polar codes to maximize mutual information(MMI)between source bits and quantized symbols.The nested structure of polar codes ensures that the MMI quantization can be implemented stage by stage.Simulation results show that the proposed MMI decoders with 4 quantization bits outperform the existing nonuniform quantized decoders that minimize mean-squared error(MMSE)with 4 quantization bits,and yield even better performance than uniform MMI quantized decoders with 5 quantization bits.Furthermore,the proposed 5-bit quantized MMI decoders approach the floating-point decoders with negligible performance loss.
基金The National Natural Science Foundation of China (No. 60496316) The National Hi-Tech Research and Development Program (863) of China (No. 2006AA01Z270)
文摘Applications on iterative control and multiple input multiple output (MIMO) system were developed. Two new charts derived from extrinsic information transfer (EXIT) chart were employed as the designing tools, which are called as output mutual information chart with defined iterative degree (DID) chart and near optimum output mutual information (NOMI) chart respectively. Different from the EXIT chart, they can show the iterative performance on the whole signal-to-noise ratio range with one single curve, whereas computation complexity is greatly reduced compared with conventional bit error ratio (BER) performance curve. The iterative control was implemented according to a near-optimum iterative degree vector determined by NOMI chart, the reasonability of uncertain parameters was analyzed in one MIMO system. The concepts were illustrated based on bit-interleaved coded modulation with iterative decoding (BICM-ID).
基金Supported by the Key Project of Chinese Ministry of Education(No.106042)the Scientific Research Foundation for the Returned Overseas Chinese Scholars,State Education Ministry(2007[24])
文摘This paper proposes a novel mapping scheme for bit-interleaved coded modulation with iterative decoding(BICM-ID).The symbol mapping is composed of two QPSK with different radiuses and phases,called cross equalization-8PSK-quasi-semi set partitioning(CE-8PSK-Quasi-SSP).Providing the same average power,the proposed scheme can increase the minimum squared Euclidean distance(MSED)and then improve the receiving performance of BICM-ID compared with conventional symbol mapping schemes.Simultaneously,a modified iteration decoding algorithm is proposed in this paper.In the process of iteration decoding,different proportion of the extrinsic information to the systematic observations results in distinct decoding performance.At high SNR(4~9dB),the observation information plays a more important role than the extrinsic information.Simulation results show that the proportion set at 1.2 is more suitable for the novel mapping in BICM-ID.When the BER is 10^(-4),more than 0.9dB coding gain over Rayleigh channels can be achieved for the improved mapping and decoding scheme.
基金supported by the National Natural Science Foundation of China(No.61205106)
文摘Considering the influence of more random atmospheric turbulence, worse pointing errors and highly dynamic link on the transmission performance of mobile multiple-input multiple-output （MIMO） free space optics （FSO） communica- tion systems, this paper establishes a channel model for the mobile platform. Based on the combination of Alamouti space-time code and time hopping ultra-wide band （TH-UWB） communications, a novel repetition space-time coding （RSTC） method for mobile 2x2 free-space optical communications with pulse position modulation （PPM） is devel- oped. In particular, two decoding methods of equal gain combining （EGC） maximum likelihood detection （MLD） and correlation matrix detection （CMD） are derived. When a quasi-static fading and weak turbulence channel model are considered, simulation results show that whether the channel state information （CSI） is known or not, the coding sys- tem demonstrates more significant performance of the symbol error rate （SER） than the uncoding. In other words, transmitting diversity can be achieved while conveying the information only through the time delays of the modulated signals transmitted from different antennas. CMD has almost the same effect of signal combining with maximal ratio combining （MRC）. However, when the channel correlation increases, SER performance of the coding 2×2 system de- grades significantly.
文摘文档分析与识别(简称文档识别)技术将各种非结构化文档数据(图像、联机笔迹)转化为结构化数据,便于计算机处理和理解,应用场景十分广阔。20世纪60年代以来,文档识别方法研究与应用受到广泛关注并取得巨大进展。得益于深度学习技术的发展和应用,文档识别的性能快速提升,相关技术在文档数字化、票据处理、笔迹录入、智能交通、文档检索与信息抽取等领域得到广泛应用。首先介绍文档识别的背景和技术范畴,回顾该领域发展历史,然后重点对深度学习方法兴起以来的研究进行综述,分析当前技术存在的不足,并建议未来值得重视的研究方向。研究现状综述部分,按文档分析与识别的几个主要技术环节(文档图像预处理、版面分析、场景文本检测、文本识别、结构化符号和图形识别、文档检索与信息抽取)分别进行介绍,简述传统方法研究的代表性工作,重点介绍深度学习方法研究的新进展。总体上,当前研究对象向深度、广度扩展,处理方法全面转向深度神经网络模型和深度学习方法,识别性能大幅提升且应用场景不断扩展。在现状分析基础上,指出当前技术在识别精度和可靠性、可解释性、学习能力和自适应性等方面还有明显不足。最后从提升性能、应用扩展、提升学习能力几个角度提出一些研究方向。从提升性能角度,研究问题包括文本识别可靠性、可解释性、全要素识别、长尾问题、多语言、复杂版面分割与理解、变形文档分析与识别等。应用扩展包括新应用(如机器人流程自动化(robotic process automation,RPA)、文字信息抄录、考古)和新技术问题(语义信息抽取、跨模态融合、面向应用的推理决策等)两方面。从提升学习能力角度,相关问题包括小样本学习、迁移学习、多任务学习、领域自适应、结构化预测、弱监督学习、自监督学习、开放集学习和跨模态学习等。
