Most recent work used raw electroencephalograph (EEG) data to train deep learning (DL) models, with the assumption that DL models can learn discriminative features by itself. It is not yet clear what kind of RSVP specific features can be selected and combined with EEG raw data to improve the RSVP classification performance of DL models. In this paper, we tried to extract RSVP specific features and combined them with EEG raw data to capture more spatial and temporal correlations of target or non-target event and improve the EEG-based RSVP target detection performance. We tested on X2 Expertise RSVP dataset to show the experiment results. We conducted detailed performance evaluations among different features and feature combinations with traditional classification models and different CNN models for within-subject and cross-subject test. Compared with state-of-the-art traditional Bagging Tree (BT) and Bayesian Linear Discriminant Analysis (BLDA) classifiers, our proposed combined features with CNN models achieved 1.1% better performance in within-subject test and 2% better performance in cross-subject test. This shed light on the ability for the combined features to be an efficient tool in RSVP target detection with deep learning models and thus improved the performance of RSVP target detection.

Multiuser massive multi-input multi-output (MU massive MIMO) is considered as a promising technology for the fifth generation (5G) of the wireless communication system. In this paper, we propose a low-complexity joint antenna and user selection scheme with block diagonalization (BD) precoding for MU massive MIMO downlink channel in the time division duplex (TDD) system. The base station (BS) is equipped with a large-scale transmit antenna array while each user is using the single receive antenna in the system. To reduce the hardware cost, BS will be implemented by limited number of radio frequency (RF) chains and BS must activate some selected transmit antennas in the BS side for data transmitting and some users' receive antennas in user side for data receiving. To achieve the reduction in the computation complexity in the antenna and user selection while maintaining the same or higher sum-rate in the system, the proposed scheme relies on three complexity reduction key factors. The first key factor is that finding the average channel gains for the transmit antenna in the BS side and the receive antenna in the user side to select the best channel gain antennas and users. The second key factor called the complexity control factor ξ(Xi) for the antenna set and the user set limitation is used to control the complexity of the brute force search. The third one is that using the assumption of the point-to-point deterministic MIMO channel model to avoid the singular value decomposition (SVD) computation in the brute force search. We show that the proposed scheme offers enormous reduction in the computation complexity while ensuring the acceptable performance in terms of total system sum-rate compared with optimal and other conventional schemes.

Since camera networks for surveillance are becoming extremely dense, finding the most informative and desirable views from different cameras are of increasing importance. In this paper, we propose a camera selection method to achieve the goal of providing the clearest visibility possible and selecting the cameras which exactly capture targets for the far-field surveillance. We design a benefit function that takes into account image visibility and the degree of target matching between different cameras. Here, visibility is defined using the entropy of intensity histogram distribution, and the target correspondence is based on activity features rather than photometric features. The proposed solution is tested in both artificial and real environments. A performance evaluation shows that our target correspondence method well suits far-field surveillance, and our proposed selection method is more effective at identifying the cameras that exactly capture the surveillance target than existing methods.

In this paper, we propose an online antenna-pulse selection method in space time adaptive processing, while maintaining considerable performance and low computational complexity. The proposed method considers the antenna-pulse selection and covariance matrix estimation at the same time by exploiting the structured clutter covariance matrix. Such prior knowledge can enhance the covariance matrix estimation accuracy and thus can provide a better objective function for antenna-pulse selection. Simulations also validate the effectiveness of the proposed method.

Activity recognition from sensors is a classification problem over time-series data. Some research in the area utilize time and frequency domain handcrafted features that differ between datasets. Another categorically different approach is to use deep learning methods for feature learning. This paper explores a middle ground in which an off-the-shelf feature extractor is used to generate a large number of candidate time-domain features followed by a feature selector that was designed to reduce the bias toward specific classification techniques. Moreover, this paper advocates the use of features that are mostly insensitive to sensor orientation and show their applicability to the activity recognition problem. The proposed approach is evaluated using six different publicly available datasets collected under various conditions using different experimental protocols and shows comparable or higher accuracy than state-of-the-art methods on most datasets but usually using an order of magnitude fewer features.

In this paper, we propose a highly efficient mobile visual search algorithm. For descriptor extraction process, we propose a low complexity feature detection which utilizes the detected local key points of the coarse octaves to guide the scale space construction and feature detection in the fine octave. The Gaussian and Laplacian operations are skipped for the unimportant area, and thus the computing time is saved. Besides, feature selection is placed before orientation computing to further reduce the complexity of feature detection by pre-discarding some unimportant local points. For the image retrieval process, we design a high-performance reranking method, which merges both the global descriptor matching score and the local descriptor similarity score (LDSS). In the calculating of LDSS, the tf-idf weighted histogram matching is performed to integrate the statistical information of the database. The results show that the proposed highly efficient approach achieves comparable performance with the state-of-the-art for mobile visual search, while the descriptor extraction complexity is largely reduced.

We propose a feature design method for a mobile visual search based on binary features and a bag-of-visual words framework. In mobile visual search, detection error and quantization error are unavoidable due to viewpoint changes and cause performance degradation. Typical approaches to visual search extract features from a single view of reference images, though such features are insufficient to manage detection and quantization errors. In this paper, we extract features from multiview synthetic images. These features are selected according to our novel reliability measure which enables robust recognition against various viewpoint changes. We regard feature selection as a maximum coverage problem. That is, we find a finite set of features maximizing an objective function under certain constraints. As this problem is NP-hard and thus computationally infeasible, we explore approximate solutions based on a greedy algorithm. For this purpose, we propose novel constraint functions which are designed to be consistent with the match conditions in the visual search method. Experiments show that the proposed method improves retrieval accuracy by 12.7 percentage points without increasing the database size or changing the search procedure. In other words, the proposed method enables more accurate search without adversely affecting the database size, computational cost, and memory requirement.

Effective capacity (EC) is an important performance metric for a time-varying wireless channel in order to evaluate the communication rate in the physical layer (PHL) while satisfying the statistical delay quality of service (QoS) requirement in data-link layer (DLL). This paper analyzes EC of amplify-and-forward wireless relay network with different relay selection (RS) protocols. First, through the analysis of the probability density function (PDF) of received signal-to-noise ratio (SNR), the exact expressions of EC for direct transmission (DT), random relay (RR), random relay with direct transmission (RR-WDT), best relay (BR) protocols are derived. Then a novel best relay with direct transmission (BR-WDT) protocol is proposed to maximize EC and an exact expression of EC for BR-WDT protocol is developed. Simulations demonstrate that the derived analytical results well match those of Monte-Carlo simulations. The proposed BR-WDT protocol can always achieve larger EC than other protocols while guaranteeing the delay QoS requirement. Moreover, the influence of distance between source and relay on EC is discussed, and optimal relay position for different RS protocols is estimated. Furthermore, EC of all protocols becomes smaller while delay QoS exponent becomes larger, and EC of BR-WDT becomes better while the number of relays becomes larger.

Ensemble learning is widely used in the field of sensor network monitoring and target identification. To improve the generalization ability and classification precision of ensemble learning, we first propose an approximate attribute reduction algorithm based on rough sets in this paper. The reduction algorithm uses mutual information to measure attribute importance and introduces a correction coefficient and an approximation parameter. Based on a random sampling strategy, we use the approximate attribute reduction algorithm to implement the multi-modal sample space perturbation. To further reduce the ensemble size and realize a dynamic subset of base classifiers that best matches the test sample, we define a similarity parameter between the test samples and training sample sets that takes the similarity and number of the training samples into consideration. We then propose a k-means clustering-based dynamic ensemble selection algorithm. Simulations show that the multi-modal perturbation method effectively selects important attributes and reduces the influence of noise on the classification results. The classification precision and runtime of experiments demonstrate the effectiveness of the proposed dynamic ensemble selection algorithm.

Sparse Matrix-Vector multiplication (SpMV) is a computational kernel widely used in many applications. Because of the importance, many different implementations have been proposed to accelerate this computational kernel. The performance characteristics of those SpMV implementations are quite different, and it is basically difficult to select the implementation that has the best performance for a given sparse matrix without performance profiling. One existing approach to the SpMV best-code selection problem is by using manually-predefined features and a machine learning model for the selection. However, it is generally hard to manually define features that can perfectly express the characteristics of the original sparse matrix necessary for the code selection. Besides, some information loss would happen by using this approach. This paper hence presents an effective deep learning mechanism for SpMV code selection best suited for a given sparse matrix. Instead of using manually-predefined features of a sparse matrix, a feature image and a deep learning network are used to map each sparse matrix to the implementation, which is expected to have the best performance, in advance of the execution. The benefits of using the proposed mechanism are discussed by calculating the prediction accuracy and the performance. According to the evaluation, the proposed mechanism can select an optimal or suboptimal implementation for an unseen sparse matrix in the test data set in most cases. These results demonstrate that, by using deep learning, a whole sparse matrix can be used to do the best implementation prediction, and the prediction accuracy achieved by the proposed mechanism is higher than that of using predefined features.

Multiuser MIMO (MU-MIMO) improves the system channel capacity by generating a large virtual MIMO channel between a base station and multiple user terminals (UTs) with effective utilization of wireless resources. Block beamforming algorithms such as Block Diagonalization (BD) and Block Maximum Signal-to-Noise ratio (BMSN) have been proposed in order to realize MU-MIMO broadcast transmission. The BD algorithm cancels inter-user interference (IUI) by creating the weights so that the channel matrices for the other users are set to be zero matrices. The BMSN algorithm has a function of maintaining a high gain response for each desired user in addition to IUI cancellation. Therefore, the BMSN algorithm generally outperforms the BD algorithm. However, when the number of transmit antennas is equal to the total number of receive antennas, the transmission rate by both BD and BMSN algorithms is decreased. This is because the eigenvalues of channel matrices are too small to support data transmission. To resolve the issue, this paper focuses on an antenna selection (AS) method at the UTs. The AS method reduces the number of pattern nulls for the other users except an intended user in the BD and BMSN algorithms. It is verified via bit error rate (BER) evaluation that the AS method is effective in the BD and BMSN algorithms, especially, when the number of user antennas with a low bit rate (i.e., low signal-to-noise power ratio) is increased. Moreover, this paper evaluates the achievable bit rate and throughput including an actual channel state information feedback based on IEEE802.11ac standard. Although the number of equivalent receive antenna is reduced to only one by the AS method when the number of antennas at the UT is two, it is shown that the throughputs by BD and BMSN with the AS method (BD-AS and BMSN-AS) are higher than those by the conventional BD and BMSN algorithms.

This paper proposes a hardware configuration for uplink multi-user multiple-input multiple-output (MU-MIMO) transmissions in a distributed antenna system (DAS). The demand for high-speed transmission in the uplink has increased recently, because of which standardizations in LTE-advanced and IEEE 802.11ax networks is currently underway. User terminal (UT) scheduling on the downlink MU-MIMO transmission is easy even in unlicensed band such as those in wireless local area network (WLAN) systems. However, the detailed management of the UTs is difficult on the uplink MU-MIMO transmissions because of the decentralized wireless access control. The proposed configuration allows an antenna to be selected from an external device on the access point (AP). All AP antennas are divided into groups, and the received signal in each group is input to the amplitude detector via a directional coupler. Subsequently, the selected antenna is fed by a multiple-to-one switch instead of a matrix switch. To clarify the effectiveness of the proposed configuration, we conduct computer simulations based on the ray-tracing method for propagation channels in an indoor environment.

Massive multiple-input multiple-output (MIMO) transmission, in which the number of antennas is considerably more than the number of user terminals, has attracted attention as a key technology in next-generation mobile communication systems, because it enables improvements in the service area and interference mitigation with simple signal processing. Multi-beam massive MIMO employing high-power beam selection in the analog part and a blind algorithm in the digital part, such as the constant modulus algorithm that does not need channel state information, has been proposed and shown to offer high transmission efficiency. In this paper, in order to realize higher transmission rates and communication efficiency, we propose a beam-selection method that uses multi-beam amplitude information only. Furthermore, this method can be realized through signal processing with a simple configuration and is highly suitable for hybrid analog-digital massive MIMO, which is advantageous in terms of cost and power consumption. Here, the effectiveness of the proposed method is verified by computer simulation.

Automatic speech recognition (ASR) and keyword search (KWS) have more and more found their way into our everyday lives, and their successes could boil down lots of factors. In these factors, large scale of speech data used for acoustic modeling is the key factor. However, it is difficult and time-consuming to acquire large scale of transcribed speech data for some languages, especially for low-resource languages. Thus, at low-resource condition, it becomes important with which transcribed data for acoustic modeling for improving the performance of ASR and KWS. In view of using acoustic data for acoustic modeling, there are two different ways. One is using the target language data, and another is using large scale of other source languages data for cross-lingual transfer. In this paper, we propose some approaches for efficient selecting acoustic data for acoustic modeling. For target language data, a submodular based unsupervised data selection approach is proposed. The submodular based unsupervised data selection could select more informative and representative utterances for manual transcription for acoustic modeling. For other source languages data, the high misclassified as target language based submodular multilingual data selection approach and knowledge based group multilingual data selection approach are proposed. When using selected multilingual data for multilingual deep neural network training for cross-lingual transfer, it could improve the performance of ASR and KWS of target language. When comparing our proposed multilingual data selection approach with language identification based multilingual data selection approach, our proposed approach also obtains better effect. In this paper, we also analyze and compare the language factor and the acoustic factor influence on the performance of ASR and KWS. The influence of different scale of target language data on the performance of ASR and KWS at mono-lingual condition and cross-lingual condition are also compared and analyzed, and some significant conclusions can be concluded.

The elastic optical network (EON) is a promising new optical technology that uses spectrum resources much more efficiently than does traditional wavelength division multiplexing (WDM). This paper focuses on the routing, modulation level, spectrum and transceiver allocation (RMSTA) problems of the EON. In contrast to previous works that consider only the routing and spectrum allocation (RSA) or routing, modulation level and spectrum allocation (RMSA) problems, we additionally consider the transceiver allocation problem. Because transceivers can be used to regenerate signals (by connecting two transceivers back-to-back) along a transmission path, different regeneration sites on a transmission path result in different spectrum and transceiver usage. Thus, the RMSTA problem is both more complex and more challenging than are the RSA and RMSA problems. To address this problem, we first propose an integer linear programming (ILP) model whose objective is to optimize the balance between spectrum usage and transceiver usage by tuning a weighting coefficient to minimize the cost of network operations. Then, we propose a novel virtual network-based heuristic algorithm to solve the problem and present the results of experiments on representative network topologies. The results verify that, compared to previous works, the proposed algorithm can significantly reduce both resource consumption and time complexity.

Public safety networks need to more effectively meet the increasing demands for images or videos to be shared among first responders and incident commanders. Long term evolution (LTE) networks are considered to be candidates to achieve such broadband services. Capital expenditures in deploying base stations need to be decreased to introduce LTE for public safety. However, out-of-coverage areas tend to occur in cell edge areas or inside buildings because the cell areas of base stations for public safety networks are larger than those for commercial networks. The 3rd Generation Partnership Program (3GPP) in Release 13 has investigated device-to-device (D2D) based relay communication as a means to fill out-of-coverage areas in public safety LTE (PS-LTE). This paper proposes a relay selection scheme based on effective path throughput from an out-of-coverage terminal to a base station via an in-coverage relay terminal, which enables the optimal relay terminal to be selected. System level simulation results assuming on radii of 20km or less revealed that the proposed scheme could provide better user ratios that satisfied the throughput requirements for video transmission than the scheme standardized in 3GPP. Additionally, an evaluation that replicates actual group of fire-fighters indicated that the proposed scheme enabled 90% of out-of-coverage users to achieve the required throughput, i.e., 1.0Mbps, to transmit video images.

This paper presents a computationally efficient cyclostationarity detection based spectrum sensing technique in cognitive radio. Traditionally, several cyclostationarity detection based spectrum sensing techniques with a low computational complexity have been presented, e.g., peak detector (PD), maximum cyclic autocorrelation selection (MCAS), and so on. PD can be affected by noise uncertainty because it requires a noise floor estimation, whereas MCAS does not require the estimation. Furthermore, the computational complexity of MCAS is greater than that of PD because MCAS must compute some statistics for signal detection instead of the estimation unnecessary whereas PD must compute only one statistic. In the presented MCAS based techniques, only one statistic must be computed. The presented technique obtains other necessary statistics from the procedure that computes the statistic. Therefore, the computational complexity of the presented is almost the same as that of PD, and it does not require the noise floor estimation for threshold. Numerical examples are shown to validate the effectiveness of the presented technique.

This paper proposes a novel method of reducing channel state information (CSI) feedback by using transmit antenna selection for downlink multiuser multiple input multiple output (DL-MU-MIMO) transmission in dense distributed antenna systems. It is widely known that DL-MU-MIMO transmission achieves higher total bit-rate by mitigating inter-user interference based on pre-coding techniques. The pre-coding techniques require CSI between access point (AP) and multiple users. However, overhead for CSI acquisition degrades the transmission efficiency of DL-MU-MIMO transmission. In the proposed CSI feedback reduction method, AP first selects the antenna set that maximizes the received power at each user, second it skips the sequence of CSI feedback for users whose signal to interference power ratio is larger than a threshold, and finally it performs DL-MU-MIMO transmission to multiple users by using the selected antenna set. To clarify the proposed method, we evaluate it by computer simulations in an indoor scenario. The results show that the proposed method can offer higher transmission efficiency than the conventional DL-MU-MIMO transmission with the usual CSI feedback method.

The main idea of filter methods in feature selection is constructing a feature-assessing criterion and searching for feature subset that optimizes the criterion. The primary principle of designing such criterion is to capture the relevance between feature subset and the class as precisely as possible. It would be difficult to compute the relevance directly due to the computation complexity when the size of feature subset grows. As a result, researchers adopt approximate strategies to measure relevance. Though these strategies worked well in some applications, they suffer from three problems: parameter determination problem, the neglect of feature interaction information and overestimation of some features. We propose a new feature selection algorithm that could compute mutual information between feature subset and the class directly without deteriorating computation complexity based on the computation of partitions. In light of the specific properties of mutual information and partitions, we propose a pruning rule and a stopping criterion to accelerate the searching speed. To evaluate the effectiveness of the proposed algorithm, we compare our algorithm to the other five algorithms in terms of the number of selected features and the classification accuracies on three classifiers. The results on the six synthetic datasets show that our algorithm performs well in capturing interaction information. The results on the thirteen real world datasets show that our algorithm selects less yet better feature subset.

The ignored side effect reflecting in the introduction of mismatching brought by contrast enhancement in representative SIFT based vein recognition model is investigated. To take advantage of contrast enhancement in increasing keypoints generation, hierarchical keypoints selection and mismatching removal strategy is designed to obtain state-of-the-art recognition result.