Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659743
Na-young Kim, Jung Kyung Lee, C. Yoo, Seunghyun Cho, Jewon Kang
In this paper, we propose a bidirectional synthesis video interpolation technique based on deep learning, using a forward and a backward video generation network and a synthesis network. The forward generation network first extrapolates a video sequence, given the past video frames, and then the backward generation network generates the same video sequence, given the future video frames. Next, a synthesis network fuses the results of the two generation networks to create an intermediate video sequence. To jointly train the video generation and synthesis networks, we define a cost function to approximate the visual quality and the motion of the interpolated video as close as possible to those of the original video. Experimental results show that the proposed technique outperforms the state-of-the art long-term video interpolation model based on deep learning.
{"title":"Video Generation and Synthesis Network for Long-term Video Interpolation","authors":"Na-young Kim, Jung Kyung Lee, C. Yoo, Seunghyun Cho, Jewon Kang","doi":"10.23919/APSIPA.2018.8659743","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659743","url":null,"abstract":"In this paper, we propose a bidirectional synthesis video interpolation technique based on deep learning, using a forward and a backward video generation network and a synthesis network. The forward generation network first extrapolates a video sequence, given the past video frames, and then the backward generation network generates the same video sequence, given the future video frames. Next, a synthesis network fuses the results of the two generation networks to create an intermediate video sequence. To jointly train the video generation and synthesis networks, we define a cost function to approximate the visual quality and the motion of the interpolated video as close as possible to those of the original video. Experimental results show that the proposed technique outperforms the state-of-the art long-term video interpolation model based on deep learning.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"86 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122100376","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper proposes speech synthesis using a WaveNet vocoder based on periodic/aperiodic decomposition. Normally, quasiperiodic and aperiodic components are contained in human speech waveforms. Therefore, it is important to accurately model periodic and aperiodic components. Periodic and aperiodic components are represented as the ratios of the energies in conventional statistical parametric speech synthesis. On the other hand, statistical parametric speech synthesis based on periodic/aperiodic decomposition has been proposed. Although the effectiveness of this approach has been shown, speech waveforms considering both periodic and aperiodic components cannot be generated directly. In this paper, we propose speech synthesis using a WaveNet vocoder based on periodic/aperiodic decomposition. In the proposed approach, separated periodic and aperiodic components are modeled by a single acoustic model based on deep neural networks, and then speech waveforms considering both periodic and aperiodic components are directly generated by a single WaveNet vocoder based on neural networks. Experimental results show that the proposed approach outperforms the conventional approach in the naturalness of the synthesized speech.
{"title":"Speech Synthesis Using WaveNet Vocoder Based on Periodic/Aperiodic Decomposition","authors":"Takato Fujimoto, Takenori Yoshimura, Kei Hashimoto, Keiichiro Oura, Yoshihiko Nankaku, K. Tokuda","doi":"10.23919/APSIPA.2018.8659541","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659541","url":null,"abstract":"This paper proposes speech synthesis using a WaveNet vocoder based on periodic/aperiodic decomposition. Normally, quasiperiodic and aperiodic components are contained in human speech waveforms. Therefore, it is important to accurately model periodic and aperiodic components. Periodic and aperiodic components are represented as the ratios of the energies in conventional statistical parametric speech synthesis. On the other hand, statistical parametric speech synthesis based on periodic/aperiodic decomposition has been proposed. Although the effectiveness of this approach has been shown, speech waveforms considering both periodic and aperiodic components cannot be generated directly. In this paper, we propose speech synthesis using a WaveNet vocoder based on periodic/aperiodic decomposition. In the proposed approach, separated periodic and aperiodic components are modeled by a single acoustic model based on deep neural networks, and then speech waveforms considering both periodic and aperiodic components are directly generated by a single WaveNet vocoder based on neural networks. Experimental results show that the proposed approach outperforms the conventional approach in the naturalness of the synthesized speech.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122212009","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A separable encryption and data insertion method is proposed in this paper. The input image is divided into 2 parts, where the first part is manipulated to mask the perceptual semantics, while the second part is processed to hide data. The binary image, which is the data to be inserted, further divides the second part of the input image into 2 regions called the ‘zero’ and ‘one’ regions. Pixels of the original image at position coinciding with the ‘zero’ region are darken, while those coinciding with the ‘one’ region are brightened. The darkening and brightening processes are performed by using histogram matching technique. The proposed joint method is separable, where the inserted binary image can be extracted directly from the masked image or from the reconstructed image. The proposed method is also commutative because the same results is achieved irregardless of the order of processing in encrypting and inserting data. Experiments were carried out to verify the basic performances of the proposed method.
{"title":"Encryption and Data Insertion Technique using Region Division and Histogram Manipulation","authors":"Ryoma Ito, Koksheik Wong, Simying Ong, Kiyoshi Tanaka","doi":"10.23919/APSIPA.2018.8659671","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659671","url":null,"abstract":"A separable encryption and data insertion method is proposed in this paper. The input image is divided into 2 parts, where the first part is manipulated to mask the perceptual semantics, while the second part is processed to hide data. The binary image, which is the data to be inserted, further divides the second part of the input image into 2 regions called the ‘zero’ and ‘one’ regions. Pixels of the original image at position coinciding with the ‘zero’ region are darken, while those coinciding with the ‘one’ region are brightened. The darkening and brightening processes are performed by using histogram matching technique. The proposed joint method is separable, where the inserted binary image can be extracted directly from the masked image or from the reconstructed image. The proposed method is also commutative because the same results is achieved irregardless of the order of processing in encrypting and inserting data. Experiments were carried out to verify the basic performances of the proposed method.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116978283","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659545
Chitralekha Gupta, Haizhou Li, Ye Wang
Automatic singing quality evaluation methods currently rely on reference singing vocals or score information for comparison. However singers may deviate from the reference singing vocal to personalize the singing that still sounds good. In this work, we present pitch histogram-based methods to automatically evaluate singing quality without any reference singing or score information. We validate the methods with the help of human ratings, and compare with the baseline methods of singing evaluation without a reference. We obtain an average Spearman's rank correlation of 0.716 with human judgments.
{"title":"Automatic Evaluation of Singing Quality without a Reference","authors":"Chitralekha Gupta, Haizhou Li, Ye Wang","doi":"10.23919/APSIPA.2018.8659545","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659545","url":null,"abstract":"Automatic singing quality evaluation methods currently rely on reference singing vocals or score information for comparison. However singers may deviate from the reference singing vocal to personalize the singing that still sounds good. In this work, we present pitch histogram-based methods to automatically evaluate singing quality without any reference singing or score information. We validate the methods with the help of human ratings, and compare with the baseline methods of singing evaluation without a reference. We obtain an average Spearman's rank correlation of 0.716 with human judgments.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128322054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659653
Qi Liu, Hui Yuan, Junhui Hou, Hao Liu, R. Hamzaoui
Rate-distortion optimal 3D point cloud compression is very challenging due to the irregular structure of 3D point clouds. For a popular 3D point cloud codec that uses octrees for geometry compression and JPEG for color compression, we first find analytical models that describe the relationship between the encoding parameters and the bitrate and distortion, respectively. We then use our models to formulate the rate-distortion optimization problem as a constrained convex optimization problem and apply an interior point method to solve it. Experimental results for six 3D point clouds show that our technique gives similar results to exhaustive search at only about 1.57% of its computational cost.
{"title":"Model-Based Encoding Parameter Optimization for 3D Point Cloud Compression","authors":"Qi Liu, Hui Yuan, Junhui Hou, Hao Liu, R. Hamzaoui","doi":"10.23919/APSIPA.2018.8659653","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659653","url":null,"abstract":"Rate-distortion optimal 3D point cloud compression is very challenging due to the irregular structure of 3D point clouds. For a popular 3D point cloud codec that uses octrees for geometry compression and JPEG for color compression, we first find analytical models that describe the relationship between the encoding parameters and the bitrate and distortion, respectively. We then use our models to formulate the rate-distortion optimization problem as a constrained convex optimization problem and apply an interior point method to solve it. Experimental results for six 3D point clouds show that our technique gives similar results to exhaustive search at only about 1.57% of its computational cost.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129689672","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659561
David B. H. Tay, Antonio Ortega, Aamir Anis
In classical multirate filter bank systems, the cascade (product) of simple polyphase matrices is an important technique for the theory, design and implementation of filter banks. A particularly important class of cascades uses elementary matrices and leads to the well known lifting scheme in wavelets. In this paper the theory and principles of cascade and lifting structures for bipartite graph filter banks are developed. Accurate spectral characterizations of these structures using equivalent subgraphs will be presented. Some features of the structures in the graph case, that are not present in the classical case, will be discussed.
{"title":"Cascade and Lifting Structures in the Spectral Domain for Bipartite Graph Filter Banks","authors":"David B. H. Tay, Antonio Ortega, Aamir Anis","doi":"10.23919/APSIPA.2018.8659561","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659561","url":null,"abstract":"In classical multirate filter bank systems, the cascade (product) of simple polyphase matrices is an important technique for the theory, design and implementation of filter banks. A particularly important class of cascades uses elementary matrices and leads to the well known lifting scheme in wavelets. In this paper the theory and principles of cascade and lifting structures for bipartite graph filter banks are developed. Accurate spectral characterizations of these structures using equivalent subgraphs will be presented. Some features of the structures in the graph case, that are not present in the classical case, will be discussed.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129881666","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659680
Kosuke Shimizu, Taizo Suzuki, K. Kameyama
This paper proposes cuboid-based perceptual encryption (Cd-PE) and a version of cube-based perceptual encryption (C-PE), named lapped cuboid-based perceptual encryption (LCd-PE), to enhance the security for Motion JPEG (MJPEG). Although C-PE provides a high level of security by dealing with several frames of the input video sequence simultaneously, keyless attackers may illegally decrypt the encrypted video sequence with conceivable attacks such as a cube-based jigsaw puzzle solver (CJPS) attack. LCd-PE subdivides the video sequence pre-encrypted with C-PE into small cuboids and further encrypts it so that attackers cannot conduct attacks such as CJPS. The experiments show that the compression performance of an encryption-then-compression (ETC) system with LCd-PE and MJPEG is almost equivalent to that of one using C-PE and yet achieves a higher level of security.
{"title":"Lapped Cuboid-based Perceptual Encryption for Motion JPEG Standard","authors":"Kosuke Shimizu, Taizo Suzuki, K. Kameyama","doi":"10.23919/APSIPA.2018.8659680","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659680","url":null,"abstract":"This paper proposes cuboid-based perceptual encryption (Cd-PE) and a version of cube-based perceptual encryption (C-PE), named lapped cuboid-based perceptual encryption (LCd-PE), to enhance the security for Motion JPEG (MJPEG). Although C-PE provides a high level of security by dealing with several frames of the input video sequence simultaneously, keyless attackers may illegally decrypt the encrypted video sequence with conceivable attacks such as a cube-based jigsaw puzzle solver (CJPS) attack. LCd-PE subdivides the video sequence pre-encrypted with C-PE into small cuboids and further encrypts it so that attackers cannot conduct attacks such as CJPS. The experiments show that the compression performance of an encryption-then-compression (ETC) system with LCd-PE and MJPEG is almost equivalent to that of one using C-PE and yet achieves a higher level of security.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130378862","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659556
Takafumi Moriya, Ryo Masumura, Taichi Asami, Yusuke Shinohara, Marc Delcroix, Y. Yamaguchi, Y. Aono
This paper presents a novel deep neural network architecture for transfer learning in acoustic models. A well-known approach for transfer leaning is using target domain data to fine-tune a pre-trained model with source model. The model is trained so as to raise its performance in the target domain. However, this approach may not fully utilize the knowledge of the pre-trained model because the pre-trained knowledge is forgotten when the target domain is updated. To solve this problem, we propose a new architecture based on progressive neural networks (PNN) that can transfer knowledge; it does not forget and can well utilize pre-trained knowledge. In addition, we introduce an enhanced PNN that uses feature augmentation to better leverage pre-trained knowledge. The proposed architecture is challenged in experiments on three different recorded Japanese speech recognition tasks (one source and two target domain tasks). In a comparison with various transfer learning approaches, our proposal achieves the lowest error rate in the target tasks.
{"title":"Progressive Neural Network-based Knowledge Transfer in Acoustic Models","authors":"Takafumi Moriya, Ryo Masumura, Taichi Asami, Yusuke Shinohara, Marc Delcroix, Y. Yamaguchi, Y. Aono","doi":"10.23919/APSIPA.2018.8659556","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659556","url":null,"abstract":"This paper presents a novel deep neural network architecture for transfer learning in acoustic models. A well-known approach for transfer leaning is using target domain data to fine-tune a pre-trained model with source model. The model is trained so as to raise its performance in the target domain. However, this approach may not fully utilize the knowledge of the pre-trained model because the pre-trained knowledge is forgotten when the target domain is updated. To solve this problem, we propose a new architecture based on progressive neural networks (PNN) that can transfer knowledge; it does not forget and can well utilize pre-trained knowledge. In addition, we introduce an enhanced PNN that uses feature augmentation to better leverage pre-trained knowledge. The proposed architecture is challenged in experiments on three different recorded Japanese speech recognition tasks (one source and two target domain tasks). In a comparison with various transfer learning approaches, our proposal achieves the lowest error rate in the target tasks.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114819494","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659784
Ryo Masumura, Suguru Kabashima, Takafumi Moriya, Satoshi Kobashikawa, Y. Yamaguchi, Y. Aono
This paper proposes relevant phonetic-aware neural acoustic models that leverage native Japanese speech and native English speech to create improved automatic speech recognition (ASR) of Japanese-English speech. In order to accurately transcribe Japanese-English speech in ASR, acoustic models are needed that are specific to Japanese-English speech since Japanese-English speech exhibits pronunciations that differ from those of native English speech. The major problem is that it is difficult to collect a lot of Japanese-English speech for constructing acoustic models. Therefore, our motivation is to efficiently leverage the significant amounts of native English and native Japanese speech material available since Japanese-English is definitely affected by both native English and native Japanese. Our idea is to utilize them indirectly to enhance the phonetic-awareness of Japanese-English acoustic models. It can be expected that the native English speech is effective in enhancing the classification performance of English-like phonemes, while the native Japanese speech is effective in enhancing the classification performance of Japanese-like phonemes. In the proposed relevant phonetic-aware neural acoustic models, this idea is implemented by utilizing bottleneck features of native English and native Japanese neural acoustic models. Our experiments construct the relevant phonetic-aware neural acoustic models by utilizing 300 hours of Japanese-English speech, 1,500 hours of native Japanese speech, and 900 hours of native English speech. We demonstrate effectiveness of our proposal using evaluation data sets that involve four levels of Japanese-English.
{"title":"Relevant Phonetic-aware Neural Acoustic Models using Native English and Japanese Speech for Japanese-English Automatic Speech Recognition","authors":"Ryo Masumura, Suguru Kabashima, Takafumi Moriya, Satoshi Kobashikawa, Y. Yamaguchi, Y. Aono","doi":"10.23919/APSIPA.2018.8659784","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659784","url":null,"abstract":"This paper proposes relevant phonetic-aware neural acoustic models that leverage native Japanese speech and native English speech to create improved automatic speech recognition (ASR) of Japanese-English speech. In order to accurately transcribe Japanese-English speech in ASR, acoustic models are needed that are specific to Japanese-English speech since Japanese-English speech exhibits pronunciations that differ from those of native English speech. The major problem is that it is difficult to collect a lot of Japanese-English speech for constructing acoustic models. Therefore, our motivation is to efficiently leverage the significant amounts of native English and native Japanese speech material available since Japanese-English is definitely affected by both native English and native Japanese. Our idea is to utilize them indirectly to enhance the phonetic-awareness of Japanese-English acoustic models. It can be expected that the native English speech is effective in enhancing the classification performance of English-like phonemes, while the native Japanese speech is effective in enhancing the classification performance of Japanese-like phonemes. In the proposed relevant phonetic-aware neural acoustic models, this idea is implemented by utilizing bottleneck features of native English and native Japanese neural acoustic models. Our experiments construct the relevant phonetic-aware neural acoustic models by utilizing 300 hours of Japanese-English speech, 1,500 hours of native Japanese speech, and 900 hours of native English speech. We demonstrate effectiveness of our proposal using evaluation data sets that involve four levels of Japanese-English.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"81 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124470498","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-11-01DOI: 10.23919/APSIPA.2018.8659531
Simon Kase, M. Tsuru, M. Uchida
It is important to observe the statistical characteristics of global flows, which are defined as series of packets between networks, for the management and operation of the Internet. However, because the Internet is a diverse and large-scale system organized by multiple distributed authorities, it is not practical (sometimes impossible) to directly measure the precise statistical characteristics of global flows. In this paper, we consider the problem of estimating the traffic rate of every unobservable global flow between corresponding origin-destination (OD) pair (hereafter referred to as “individual-flows”) based on the measured data of aggregated traffic rates of individual flows (hereafter referred to as “aggregated-flows”), which can be easily measured at certain links (e.g., router interfaces) in a network. In order to solve the OD traffic matrix estimation problem, the prior method uses an inverse function mapping from the probability distributions of the traffic rate of aggregated-flows to those of individual-flows. However, because this inverse function method is executed recursively, the accuracy of estimation is heavily affected by the initial values of recursion and variation of the measurement data. In order to solve this issue and improve estimation accuracy, we propose a method based on a resampling of measurement data to obtain a set of solution candidates for OD traffic matrix estimation. The results of performance evaluations using a real traffic trace demonstrate that the proposed method achieves better estimation accuracy than the prior method.
{"title":"Accurate OD Traffic Matrix Estimation Based on Resampling of Observed Flow Data","authors":"Simon Kase, M. Tsuru, M. Uchida","doi":"10.23919/APSIPA.2018.8659531","DOIUrl":"https://doi.org/10.23919/APSIPA.2018.8659531","url":null,"abstract":"It is important to observe the statistical characteristics of global flows, which are defined as series of packets between networks, for the management and operation of the Internet. However, because the Internet is a diverse and large-scale system organized by multiple distributed authorities, it is not practical (sometimes impossible) to directly measure the precise statistical characteristics of global flows. In this paper, we consider the problem of estimating the traffic rate of every unobservable global flow between corresponding origin-destination (OD) pair (hereafter referred to as “individual-flows”) based on the measured data of aggregated traffic rates of individual flows (hereafter referred to as “aggregated-flows”), which can be easily measured at certain links (e.g., router interfaces) in a network. In order to solve the OD traffic matrix estimation problem, the prior method uses an inverse function mapping from the probability distributions of the traffic rate of aggregated-flows to those of individual-flows. However, because this inverse function method is executed recursively, the accuracy of estimation is heavily affected by the initial values of recursion and variation of the measurement data. In order to solve this issue and improve estimation accuracy, we propose a method based on a resampling of measurement data to obtain a set of solution candidates for OD traffic matrix estimation. The results of performance evaluations using a real traffic trace demonstrate that the proposed method achieves better estimation accuracy than the prior method.","PeriodicalId":287799,"journal":{"name":"2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124527528","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：