Search

Voice activity detection (VAD) is essential for multiple microphone arrays processing, in which massive potential devices, such as microphone devices for far-field voice-based interaction in smart home environments, will be activated when sound sources appear. Therefore, the VAD can save a lot of computing resources in massive microphone arrays processing for the sparsity in sound source activity. However, it may not be feasible to obtain an accurate VAD in harsh environments, such as far-field, time-varying noise field. In this paper, the long-term speech information (LTSI) and the log-energy are modeled for deriving a more accurate VAD. First, the LTSI can be obtained by measuring the differential entropy of long-term smoothed noisy signal spectrum. Then, the LTSI is used to get labeled data for the initialization of a Gaussian mixture model (GMM), which is used to fit the log-energy distribution of noise and (noisy) speech. Finally, combining the LTSI and the GMM parameters of noise and speech distribution, this paper derives an adaptive threshold, which represents a reasonable boundary between noise and speech. Experimental results show that our VAD method has a remarkable improvement for a massive microphone network.

This paper presents a method to deploy the virtual router on the network node of OpenStack platform by using the networking VPP (Vector Packet Process) as the virtual router. This method is dedicated to high-speed network tasks. It runs on the DPDK (Data Plane Development Kit) user space, does not use the kernel driver to obtain packets from the hardware. The sending, receiving and processing of messages are implemented in user space, which means less kernel calls, less context switching and faster processing. In this method, the computing nodes and network nodes are deployed separately, and the user mode forwarding work of network node data plane is transferred to VPP. VPP is used to realize the function of three-layer routing and forwarding, which can accelerate the east-west traffic of VLAN network, also can improve the performance of three-layer routing and forwarding by 3–5 times. By this method, the bottleneck of OpenStack network can be eliminated, the performance of data center network can be further optimized.

This paper investigates the physical layer security of non-orthogonal multiple access (NOMA) with decode-and-forward (DF) relays in underlay cognitive radio (CR) network. Secondary transmitter communicates with legitimate users via the aid of relays, while the eavesdroppers try to intercept the message. To safeguard legitimate communications against eavesdropping, the joint of artificial-noise (AN) and relay selection (RS) is applied. We derive the secrecy outage probability (SOP) and the secrecy throughput to evaluate the secrecy performance. In addition, the asymptotic floor of SOP is derived. Numerical results verify that cooperative NOMA in underlay CR (CRNOMA) significantly outperforms the conventional orthogonal multiple access (OMA) in underlay CR network (CROMA) in terms of the secrecy outage performance. Furthermore, the effect of some critical parameters on the secrecy performance has been studied.

© 2020 Elsevier Inc. Using barycentric coordinates for thickness interpolation, we present a novel polygonal skeleton based convolution surface approximation method with varying radii. Given the prescribed radii of an arbitrary planar polygonal skeleton, we first employ a smooth interior mean value coordinate interpolation approach to calculate the thickness at each projection position in the polygonal plane. Then a local thickness approximation method based on finite-support kernels is introduced to create an implicit surface with smoothly varying thickness. In addition, a polygon offset with different distances by winding numbers is employed to create local approximation at polygon boundaries. Our experiments show that the proposed uniform smooth thickness interpolation and local convolution approximation method can not only avoid surface wrinkles but also reduce computation cost. Moreover, our approach is insensitive to exterior thickness interpolation. Therefore smooth barycentric coordinates within a polygon can all be easily integrated into our approach.

We investigate the problem of target tracking using a wireless sensor network with asynchronous sensors. To study the impact of sensor clock imperfection on target tracking in practical situations, we build a testbed and collect data from an outdoor experiment. After analyzing the collected data, we find that the TDOA (time-difference-of-arrival) and FDOA (frequency-difference-of-arrival) measurements have notable bias, which is caused by asynchronous sensors or more precisely by the sensor clock drift. Based on the model of clock drift, the measurement bias and the target position are integrated into a state-space model. Both can be estimated in the framework of the extend Kalman filter. In some circumstance, the target trajectory is tracked successfully.

Although multimedia surveillance systems are becoming increasingly ubiquitous in our living environment, automated multimedia surveillance systems based on video camera lacks the robustness and reliability most of the time in several real applications. To overcome this drawback, audio sensory devices have been taken into account in a considerable amount of research. For example, Sound Source Localization (SSL) may indicate potential security risks and could point the camera in that direction. In this paper, a reliable sound source localization based on Time-Difference-Of-Arrival (TDOA) is explored. The novel aspect of our approach includes a TDOA based Gaussian filter to improve the accuracy and stability of sound source localization. The advantage of our proposed algorithm is its extensive integration with various TDOA-based methods in all kinds of microphone array. The Experimental comparison shows significant improvement over the state of the art TDOA-based algorithm.

© 2019 John Wiley & Sons, Ltd. Although different approaches of segmenting and assembling geometric models for 3D printing have been proposed, it is difficult to find any research studies, which investigate model segmentation and assembly in head-mounted display (HMD) virtual reality (VR) environments for 3D printing. In this work, we propose a novel and interactive segmentation method for screwing assembly in the environments to tackle this problem. Our approach divides a large model into semantic parts with a screwing interface for repeated tight assembly. Specifically, after a user places the cutting interface, our algorithm computes the bounding box of the current part automatically for subsequent multicomponent semantic Boolean segmentations. Afterwards, the bolt is positioned with an improved K3M image thinning algorithm and is used for merging paired components with union and subtraction Boolean operations respectively. Moreover, we introduce a swept Boolean-based rotation collision detection and location method to guarantee a collision-free screwing assembly. Experiments show that our approach provides a new interactive multicomponent semantic segmentation tool that supports not only repeated installation and disassembly but also tight and aligned assembly.

State-of-the-art piecewise logarithmic converters employ nonuniform segments to reduce the variance of relative errors, particularly for small inputs. However, it is difficult to have an optimized algorithm to accurately choose the nonuniform segments for a required approximation error. In this brief, we present a relative error equal distribution (REED) algorithm that performs the nonuniform piecewise linear interpolation of logarithm. It is able to precisely set the nonuniform piecewise points so that the relative error of each segment is within an upper bound, and finally produce a flat relative error distribution. It can be applied to any number of segments, even if the number is not the power of 2. Experimental results show that our REED algorithm achieves over 70% reduction of average relative errors for the benchmark graphics application, compared to both the state-of-the-art uniform and nonuniform methods. Synthesis results show that our hardware resources are comparable to that of the state-of-the-art nonuniform methods.

In this paper, we propose an end to end model for keyword spotting(KWS) with densely connected convolutional network(DenseNet) and the integration of Short Time Fourier Transform(STFT) which aims to extract utterances from raw waveform directly. Furthermore, we investigate the efficiency of adaptive and trainable window function in the task of keywords spotting. Using the recently-released Google Speech Commands Dataset as our benchmark. Our DenseNet implementation significantly outperforms previous neural networks based KWS system in terms of accuracy.

Voice activity detection (VAD) is a very challenging problem in adverse acoustic environments (e.g. far-field and conditions with different types of noise). In this paper, we proposed a Gaussian mixture model (GMM) for log-energy distribution of noise and (noisy) speech, where the distribution of these two components can be self-adapting in non-stationary circumstances. An adaptive threshold based on the GMM parameters of these two components represents a reasonable bound between noise and speech, which can lead to an accurate VAD in various noise conditions. To further improve speech hit rate (SHR) and non-speech hit rate (NHR), some constraints are introduced to this proposed GMM for reliability. Experimental results demonstrate that the proposed method yields remarkable performance for SHR and NHR.

Hierarchy is a fundamental characteristic of many complex systems. The methods of structural information have been taken as a prospective way for quantifying dynamical network complexity. This paper is based on the study of the high-dimensional natural structural information entropy in networks. And then we propose a new similarity District Structural Information (DSI) index, which takes the characteristics of network districts into consideration, to analyze the complexity of dynamical network districts. Based on the method, this paper applies the district structural information to explain the equilibrium problem in real-world networks. And taking Beijing traffic network and its districts to complete experiments demonstrates that the DSI index can reflect the equilibrium of the network and the districts effectively.

Hierarchy is a fundamental characteristic of many complex systems. The methods of structural information have been taken as a prospective way for quantifying dynamical network complexity. This paper based on the study of the high-dimensional natural structural information entropy in networks. And then propose a new similarity District Structural Information (DSI) index, which takes the characteristics of network districts into consideration, to analyze the complexity of dynamical network districts. Based on the method, this paper apply the district structural information to explain the equilibrium problem in real-world networks. And take the Beijing traffic network and its districts to complete experiments that demonstrate the DSI index can reflect the equilibrium of the network and the districts effectively.

We propose an end-to-end speaker verification system based on the neural network and trained by a loss function with less computational complexity. The end-to-end speaker verification system in this paper consists of a ResNet architecture to extract features from utterance, then produces utterance-level speaker embeddings, and train using the large-margin Gaussian Mixture loss function. Influenced by the large-margin and likelihood regularization, large-margin Gaussian Mixture loss function benefits the speaker verification performance. Experimental results demonstrate that the Residual CNN with large-margin Gaussian Mixture loss outperforms DNN-based i-vector baseline by more than 10% improvement in accuracy rate., Comment: 5 pages. arXiv admin note: text overlap with arXiv:1803.02988, arXiv:1705.02304, arXiv:1706.08612 by other authors

International audience; The standard approaches of road traffic flow measurement as a part of advanced traffic management system relies on data acquisition from inductive loops or visual detectors. Due to their high cost and a number of operational limitations, this study was to elaborate a new concept of traffic flow estimation based on data from acoustic sensors. The experimental study has been conducted on a roadside of Paris ring-road (peripherique boulevard) during 11.5 days. The obtained data has been processed with help of Support Vector Regression method. The performances of the proposed solution have been assessed against standard traffic flow measurements. Obtained result show that this approach is promising and has potential of usage as independent measurement system and as auxiliary unit for existing systems.

With the rapid development of location-based services in the field of mobile network applications, users enjoy the convenience of location-based services on one side, while being exposed to the risk of disclosure of privacy on the other side. Attacker will make a fierce attack based on the probability of inquiry, map data, point of interest (POI), and other supplementary information. The existing location privacy protection techniques seldom consider the supplementary information held by attackers and usually only generate single cloaking region according to the protected location point, and the query efficiency is relatively low. In this paper, we improve the existing LBSs system framework, in which we generate double cloaking regions by constraining the supplementary information, and then k-anonymous task is achieved by the cooperation of the double cloaking regions; specifically speaking, k dummy points of fixed dummy positions in the double cloaking regions are generated and the LBSs query is then performed. Finally, the effectiveness of the proposed method is verified by the experiments on real datasets.

Easy and efficient 3D modelling in virtual environments is an important and unsolved topic. This paper proposes a new modelling approach to tackle this issue. It invents convolution surface-based brushes to directly draw 3D models in head-mounted display-based virtual environments. In order to maximize the efficiency, flexibility, and capacity of our proposed modelling approach, we propose three different skeleton-based convolution surfaces to tackle different modelling tasks: point skeleton-based convolution surfaces for metaball shapes, line skeleton-based convolution surfaces for cylindrical shapes, and polygon skeleton-based convolution surfaces for planar surfaces. Their combination makes 3D modelling more flexible and powerful. The high efficiency is further raised by our developed closed-form solutions for point skeletons, ends of line skeletons, and edges of polygonal skeletons. Different user-friendly sweeping schemes are provided to facilitate intuitive inputs for various complex shape generation. Unlike Google's Tilt Brush, which is used to create disconnected sheet-like surfaces only, our proposed convolution surface-based brushes can produce smoothly blended manifold surfaces, and novice users can easily learn and use them to create various interesting 3D models efficiently.