Machine Learning and Intelligent Communications

In mobile ad-hoc networks, the random mobility of nodes will result in unreliable connection. In addition, the bandwidth resource limit will affect the quality of service (QoS) critically. In this paper, an effective QoS-based reliable route selecting scheme (QRRSS) is proposed to alleviate the above problems. The route reliability can be estimated by received signal strength and the control packet overhead can be decreased by selecting more reliable link that satisfies the QoS requirements. Simulation results indicate that the reliable route selecting scheme presented in this paper shows obvious superiority to the traditional ad-hoc QoS on-demand routing (AQOR) in the packet successful delivery rate, the control packet overhead and the average end-to-end delay.

This paper presents a distributed, compressive multiple target localization and tracking system based on wireless fiber-optic sensors. This research aims to develop a novel, efficient, low data-throughput multiple target tracking platform. The platform is developed based on three main technologies: (1) multiplex sensing, (2) space encoding and (3) compressive localization. Multiplex sensing is adopted to enhance sensing efficiency. Space encoding can convert the location information of multi-target into a set of codes. Compressive localization further reduces the number of sensors and data-throughput. In this work, a graphical model is employed to model the variables and parameters of this tracking system, and tracking is implemented through an Expectation-Maximization (EM) procedure. The results demonstrated that the proposed system is efficient in multi-target tracking.

In the field of image processing, Gaussian mixture model (GMM) is always used to detect and recognize moving objects. Due to the defects of GMM, there are some error detections in the final consequence. In order to eliminate the defects of GMM in moving objects detections, this paper has studied a moving object detection algorithm, combining GMM with scale-invariant feature transform (SIFT) keypoint match. First, GMM is built to obtain the distributions of background image pixels. Then, morphological processing method is applied to improve the quality of binary segmentation image and extract segmentation images of moving objects. Finally, SIFT keypoint match algorithm is used to eliminate misjudgment segmentation images by judging whether the segmentation image matches with the background template or not. Compared with original GMM, the results show that the accuracy of moving object detection has been improved.

This paper proposes low-complexity detection algorithms for Massive MIMO system: Multiple Dominant Eigenvector Detection Algorithm (MDEDA) and Antenna Selection Scheme (ASS). Both the schemes calculate the log likelihood ratios (LLRs). Based on the Single Dominant Eigenvector Detection (SDEDA), MDEDA searches transmitted signal candidates in multiple dominant eigenvector directions. For one thing, combined multiple eigenvectors, MDEDA attains better BER performance, for another, it greatly reduces the number of transmitted signal candidates. The ASS contains Single Antenna Selection Scheme (SASS) and Multiple Antenna Selection Scheme (MASS), focus on channel error modeling, the ASS assumes the signal of some antennas corresponding to the constellation points in order to minimize the channel error. SASS searches all transmit antennas, nevertheless, MASS chooses multiple antennas based on the eigenvalue. Finally, SASS gains better BER performance but more complexity. Finally, SASS provides an excellent trade-off between performance and complexity.

In recent years, several correlation tracking algorithms have been proposed exploiting hierarchical features from deep convolutional neural networks. However, most of these methods focus on utilizing the CNN features for target location and neglect the changes of target scale, which may import error to the model and lead to drifting. In this paper, we propose a novel scale-variable tracking algorithm based on hierarchical CNN features, which learns correlation filters to locate the target and constructs a target pyramid for scale estimation. To evaluate the tracking algorithm, extensive experiments are conducted on a benchmark with 100 video sequences, which demonstrate features exploited from different CNN layers are well fit to estimate the object scale. The evaluation results show that our tracker outperforms the state-of-the-art methods by a huge margin (+14.6% mean OS rate and +14.3% mean DP rate).

As a high-resolution deep tissue imaging technology, photoacoustic microscopy (PAM) is attracting extensive attention in biomedical studies. PAM has trouble in achieving real-time imaging with the long data acquisition time caused by point-to-point sample mode. In this paper, we propose a sparse photoacoustic microscopy reconstruction method based on matrix nuclear norm minimization. We use random sparse sampling instead of traditional full sampling and regard the sparse PAM reconstruction problem as a nuclear norm minimization problem, which is efficiently solved under alternating direction method of multiplier (ADMM) framework. Results from PAM experiments indicate the proposed method could work well in fast imaging. The proposed method is also be expected to promote the achievement of PAM real-time imaging.

Image segmentation plays an important role in the field of digital production management. Image resolution is an important factor affecting the size of its segmentation and segmentation efficiency, and the physical characteristics of the image capturing device is another important factor. With high-resolution segmentation algorithm in image segmentation, we often find that the edge contour image segmentation is difficult to accurately determine, more complex image arithmetic operation efficiency is not high and images taken with a different device in response to segmentation algorithms are very different. In this paper, the plant leaf image collected from different cameras was used as the object of study, and the feature quantity was extracted. The appropriate segmentation boundary was determined by cluster analysis. The leaf image was pretreated with the resolution adjustment, and the leaf image was in the appropriate segmentation feature range. After the clustering domain processing of the feature range in this paper, it solves the problem that the real edge of the leaf area information is too difficult to distinguish, and effectively solves the problem of complex image algorithm and ordinary pc machine in the process of complex image processing Efficiency issues. The appropriate segmentation feature range of the devices is established for different devices, which effectively solves the different response of different devices to the segmentation algorithm.

Traditional semi-blind channel estimator is based on eigen value decomposition (EVD) or singular value decomposition (SVD), which effectively reduces the interference through dividing the observed signal into signal subspace and noise subspace. Due to the large computation, Massive MIMO systems could not afford the cost of traditional algorithms in spite of the high performance. In this paper, we propose a channel estimation algorithm based on subspace tracking, in which the signal subspace is obtained by approximating power iteration algorithm. Without sacrificing the estimation performance, the complexity is greatly reduced compared with the traditional semi-blind channel estimation algorithm, which improves the applicability of the estimator.

Downlink multi-user shared access (MUSA) is a non-orthogonal multiple access scheme (NOMA) based on the traditional power domain superposition and uses a mirror constellation to optimize the modulated symbol mapping of the paired users. In this paper, bit error ratio (BER) performance of MUSA with successive interference cancellation (SIC) is investigated in a cellular downlink scenario over Rayleigh fading channel. Firstly, we elaborate downlink MUSA system based on NOMA and spreading sequences in detail. Then, we compare the BER performance of MUSA with pure NOMA under different power allocation schemes. On this basis, we further study the system average BER performance in downlink MUSA and NOMA with respect to the power difference of the users, respectively. In addition, BER performance of MUSA with different spreading sequences is evaluated. Finally, the simulation results show that MUSA with appropriate spreading sequences is able to obtain better BER performance than NOMA under the same simulation conditions, and a reasonable power allocation is the key to improve BER performance of MUSA and NOMA.

Recently, there is an increase interest in location sharing services in social networks. Behind the convenience brought by location sharing, there comes an indispensable security risk of privacy. Though many efforts have been made to protect user’s privacy for location sharing, they are not suitable for social network. Most importantly, little research so far can support user relationship privacy and identity privacy. Thus, we propose a new privacy protection protocol for location sharing in social networks. Different from previous work, the proposed protocol can provide perfect privacy for location sharing services. Simulation results validate the feasibility and efficiency of the proposed protocol.

Wireless localization has become a key technology location based services, and the non-line-of-sight (NLOS) propagation is one of the most important error source in the localization. Therefore, this paper defines a novel algorithm residual error (ARE) in NOLS environment, and estimates the position of mobile station (MS) by minimizing this ARE, where the quadratic programming is employed to solve the minimization problem. The simulation results show that the proposed algorithm produces significant performance improvements in NLOS environments.

With the development of information technology and the rising of demanding for location-based services, indoor localization has obtained great attentions. Accurate estimation of Angle of Arrival (AoA) of signals make it possible to achieve a high precision location. So as to resolve multipath signals effectively and then extract AoA of the direct path, in this paper we first use the existing three-antenna commercial Wi-Fi Network Interface Card (NIC) to collect radio Channel Frequency Response (CFR) measurements and then jointly estimate AoA and Time of Arrival (ToA). Second, we propose a sensing algorithm to distinguish Line-of-Sight (LoS) and Non-Line-of-Sight (NLoS) propagation and therefore obtain finer localization. Our experiments in a rich multipath indoor environment show that the AoA-based the proposed localization system can achieve a median accuracy of 0.8 m and 1.3 m in LoS environment and NLoS environment, respectively.

Deep learning is employed to detect defects in photovoltaic (PV) modules in the thesis. Firstly, the thesis introduces related concepts of cracks. Then a convolutional neural network with seven layers is constructed to classify the defective battery panels. Finally, the accuracy of the validation set is 98.35%. Besides, the thesis introduces a method in which a single battery cell can be extracted from the Electro Luminescence (EL) image of the PV module. This method is very suitable for automatic inspection of photovoltaic power plants.

In indoor positioning system based on fingerprint, the traditional fingerprint database construction method consumes much manpower and time cost. To solve this problem, we propose an effective method for constructing fingerprint database by using Microelectro Mechanical System (MEMS) to assist Bluetooth Low Energy (BLE), which overcomes the low efficiency of traditional methods. Meanwhile, the method achieves the comparable positioning accuracy and reduces workload more than 70%. In the optimization procedure, we use affine propagation clustering, outlier detection and filtering of Received Signal Strength Indication (RSSI) to optimize fingerprint database. Finally, the BLE positioning error conducted by the effective database is about 2 m.

The JPEG (Joint Photographic Experts Group) file format is currently one of the most widely used image formats. The study on JPEG steganography and steganalysis is a hotspot in the field of information hiding. With the matrix coding and some new adaptive embedding strategies having been put forward, the detection of stego images is becoming more and more difficult. In recent years, a series of new feature extraction methods have been proposed in the field of steganalysis. However, the detection accuracy rate can only be increased by 1–2% points or even less. Based on those existing steganalytic algorithms, a new feature merging method is proposed in this paper. Via merging features extracted from different domains, the detection accuracy rate of those existing JPEG steganalytic algorithms can be improved by 3% points or even higher. Considering about that the feature dimension is so high after feature merging and thus it may bring difficulties in the feature extraction, training and classification process, a new feature selection method is also proposed in this paper. Experimental results demonstrate that it can not only achieve reduction of the dimensionality, but also maintain a high detection accuracy rate.

Detection of double JPEG compression plays an increasingly important role in image forensics. This paper mainly focuses on the situation where the images are aligned double JPEG compressed with two different quantization tables. We propose a new detection method based on the fusion features of Benford features and likelihood probability ratio features in this paper. We believe that with the help of likelihood probability ratio features, our fusion features can expose more artifacts left by double JPEG compression, which lead to a better performance. Comparative experiments have been carried out in our paper, and experimental result shows our method outperforms the baseline methods, even when one of the quality factors is pretty high.

Sensor patter noise (SPN) has been proved to be an unique fingerprint of a camera, and widely used for camera source identification. Previous works mostly construct reference SPN by averaging the noise residuals extracted from images like blue sky. However, this is unrealistic in practice and the noise residual would be seriously affected by scene detail, which would significantly influence the performance of camera source identification. To address this problem, a complexity based sample selection method is proposed in this paper. The proposed method is adopted before the extraction of noise residual to select image patches with less scene detail to generate the reference SPN. An extensive comparative experiments show its effectiveness in eliminating the influence of image content and improving the identification accuracy of the existing methods.

For reducing the complexity of equalization, linear equalization can be adopted for generalized spatial modulation (GSM) which is a special case of multiple-input-and-multiple-output (MIMO). However, because of its inferior performance, linear equalization may be infeasible for practical GSM systems which has large number of antennas and constellation. On the other hand, lattice-reduction (LR) is an effective method to improve the performance of linear equalization. The lattice reduction can’t be utilized by GSM directly, because signals on some antennas don’t exist. For tackling this problem, we propose a compatible 8-QAM constellation scheme integrating LR-aided linear equalization with GSM effectively. Next, we prove that LR-aided linear equalizers collect the same diversity order as that exploited by the ML detector under Rayleigh fading channels, and implement some simulations. Simulation results show the superior of the proposed 8-QAM over traditional 4-QAM and 8-QAM under Rayleigh fading channel. Moreover, our scheme obtains the full receive diversity under correlated channel.

The Radio frequency fingerprinting (RFF) generation mechanism is analyzed in this paper. It is proved to be a secure means for network security access. At the same time, the method of RFF extraction is also given. The characteristics of RFF are analyzed theoretically. Then, a high-precision fingerprint feature identification method based on Kalman filter is proposed. The results of the experiments show that the proposed system can work effectively in the environment where the signal-to-noise ratio (SNR) is higher than 10 dB, and the achieved identification rate is higher than 90%.

In the mobile network, building a prediction based network traffic model is of great significance for mobile network optimization, so that the operators is able to schedule the resources adaptively. In the paper, multiplicative seasonal Autoregressive Integrated Moving Average model (ARIMA) and Holt-Winters model are proposed for modeling of traffic predication, where the historical traffic series of a typical tourist area are utilized to verify the performance. The two methods analyze the trend of mobile network traffic per hour, build and validate models. Then predict mobile network traffic within a given period of time. The error rate of different models predictions is analyzed to provide certain decision basis for the allocation of network resources.

Aiming at the problem of side obstacles on the transmission path, an algorithm for more accurate prediction of field strength has been proposed. The algorithm uses the slant profile which is determined by the antenna line and the minimum Fresnel radius to gather the information of the side obstacles, determine the Fresnel clearance and calculate the diffraction loss. The simulation results show that the side obstacles can produce the diffraction loss in the radio wave propagation as same as the vertical obstacle can, and verify the correctness and rationality of the attenuation calculated by the method in the case of the side obstacle.

Signal acquisition in ultra-high frequency is a challenging problem due to high cost of analog-digital converter. While compressed sensing (CS) provides an alternative way to sample signal with low sampling rate, the construction of measurement matrix is still challenging due to hardware complexity and random generation. To address this challenge, a variable dimension deterministic measurement matrix construction method is proposed in this paper based on cross-correlation characteristics of m sequences. Specifically, a lower bound of the spark of measurement matrix is derived theoretically. The proposed measurement matrix construction method is applicable to compressive sampling system to improve the quality of signal reconstruction, especially for modulated wideband converter (MWC) architecture. Simulation results demonstrate that the proposed measurement matrix is superior to random Gauss matrix and random Bernoulli matrix.

Detecting signal interference and assessing signal quality are essential tasks to ensure the normal communication within an area. As for traditional methods, we have to take field measurements after setting up a base station which needs to obtain huge data in low efficiency. Aiming at this particular problem, this paper proposed to assess signal quality by compressed sensing. Method of compressed sensing used in signal quality assessment is firstly discussed. After that, we introduced the specific process when assessing. At last the results of reconstructing the measured data and the predicted data separately shows that it could met the accuracy requirements of signal quality assessment.

For multi-antenna system, the difficulties of preforming spectrum sensing are high sampling rate and hardware cost. To alleviate these problems, we propose a novel utilization of distributed compressive sensing for the multi-antenna case. The multi-antenna signals first are sampled in terms of distributed compressive sensing, and then the time-domain signals are reconstructed. Finally, spectrum sensing is performed with help of energy-based sensing method. To evaluate the proposed method, we do the corresponding simulations. The simulation results proves the proposed method.

Electromagnetic spectrum, the main medium of wireless communication has been over-crowded. Accompanied by the arrival of big data era, the problem of the spectrum scarcity has received people’s attention. The emergence of cognitive radio improves the utilization of the spectrum and provides an effective solution to break the limitations of the traditional static allocation. Radio Environmental Maps (REM) is an enabling technology of cognitive radio which can be intuitive, multi-dimensional display of spectrum information. It provides a visual basis while accessing dynamic spectrum and sharing spectrum. In this paper, the various aspects of REM are studied from the perspective of cognitive radio. Based on the concept of REM, the recent research progress of REM is summarized, and a series of challenges in the construction of spectrum pattern are also highlighted.

In order to eliminate downlink interference and improve system performance, we proposed a method to eliminate inter-distract downlink interference based on the non-overlapping nature of the signal in autocorrelation domain. In this method, multi-antenna technology was used and spectrum resource was not additionally occupied, without requiring channel conditions. The simulation result showed that this method is suitable for the removal of strong downlink interference in the mobile station at the edge of the distract.

Wireless networks have been widely used in various industries, e.g., the subway communication system. However, there are many technical problems that are still unsolved. One of the most important issues about the problem is the reliability of “train-to-earth” communication. Therefore, to enhance the stability and reliability of the subway wireless communication system, we dispose the dual gateways and stations, and propose a dual-stations collaborative communication scheme. Implementation between dual gateways and dual stations are required to design a state testing scheme. Therefore, the system reliability is improved as it can detect the system malfunctions. There are three kinds of work modes where each mode has been designed to solve the stated problems previously. The experiment results show that the proposed scheme can reach the expected requirements, thus achieving the reliable and secure “train-to-earth” communication.

A direction recognition attendance system based on RFID (Radio Frequency Identification) is designed in the paper. Using multiple card readers (a master and more slave), the system can recognize the direction of the cardholders effectively. Firstly, to read the RFID cards held by passersby and vehicles, multiple card readers must be installed in the region. Secondly, according to reading the difference of recorded time by multiple card readers, the direction of passage can be decided. Synchronism of the master-slave card readers are achieved using the time hack command, which ensure the accuracy of the decided direction. Finally, the access records will be packaged and transmitted to the server by the mobile network from the master card reader. The system can decide the direction of passage and calculate the passing time of the passersby and vehicles, making it a highly intelligent and efficient attendance management system.

As the development of economy and industry, air quality decreases as one of the exchanges of our achievements. Although air pollution has already been considered as a global and critical issue over the past decades, there has not been much innovation on the way people monitor and check the quality. Most of the air quality data today is provided by government or professional sensors set up in cities, which does not provide more detailed status in smaller geo locations with finer granularity, such as specific villages, schools, and shopping malls. In this project, we use machine learning to make a mathematical model which could be used to predict the air quality for small geo locations with accuracy and fine granularity. Through series of experiments and comparisons, the most accuracy mathematical model was found, which had a difference percentage less than 20% with the real data.

This paper studies the combination of information network technologies like Internet of Things (IoT) and big data with traditional Chinese medicine (TCM) to build a system framework oriented to TCM smart service. TCM-oriented knowledge representation technology is also explored so as to realize computer recognition and calculation of TCM health service, the self-learning reasoning technology of system is further studied, and TCM knowledge fuzzy model and modified BP neural network algorithm are introduced into TCM smart service system to conduct machine learning and smart judgment upon various diseases. These technologies will promote the scientific research and artificial intelligence aided diagnosis of TCM.

The development of technology in large strides has enabled wireless sensor network to extensively supersede traditional wired sensor network (WSN), which is accompanied with its application to every aspect of life and production. In terms of smart buildings, presently the mainstream direction in the research is combining wireless sensor network with IOT technology and internet technology, etc. In this paper, based on ZigBee wireless sensor network (WSN) combined with Java, Android, etc., developed to monitor building real-time environmental data of intelligent building system, and can achieve the combination of automatic and manual household appliance control platform.

Distributed system is more complicated compared with other systems due to its concurrency and distribution. Moreover, the traditional system development process is usually informal, and a large number of tests are required. On the other hand, the formal methods have been applied in many system development fields and many achievements have been made. In this paper, a method which combines SysML requirement diagrams and Event-B to model distributed system is proposed, including their mapping relations.

With the rapid development of space networks, new space communications protocols are emerging, for which emulation is an essential step during design and test. In this paper, we propose a lab-based testbed, in which software and hardware tools are utilized together to emulate full network protocol stack. A software protocol gateway is implemented to preform protocol conversion like IP over CCSDS in Data Link Layer. A specified hardware, Cortex CRT-Q is adopted for accurate emulation of space links, which connects upper layers with Physical Layer. Thus, our testbed benefit from both the fidelity provided by hardware and flexibility brought by software.

Space communications have the characteristics of long link delays, frequent link disruptions and high error rates. With reliable Licklider Transmission Protocol (LTP) or Transmission Control Protocol (TCP), automatic repeat request (ARQ) is applied to enable reliable data delivery in delay/disruption tolerant networking (DTN). However, ARQ is inefficient for space communications especially in links with long round trip time (RTT). In this paper, an application layer Reed-Solomon (ALRS) channel coding scheme is proposed, which is further combined with ARQ to guarantee reliable transmission in DTN architecture. The proposed ALRS coding scheme is implemented in open source ION-DTN software and its performance is evaluated on a dedicated testbed. The results of the experiments show that this scheme in DTN can be speed up in most scenarios compared with ARQ-only scheme. With coding in application layer, the scheme is also more compatible with the overlay characteristic of DTN.

As the scale of small satellite network is not large and the transmission cost is high, it is necessary to optimize the routing problem. We apply the traditional time-expanded graph to model the data acquisition of small satellite network so that we can formulate the data acquisition into a multi-commodity concurrent flow optimization problem (MCFP) aiming at maximizing the throughput. We use an approximation method to accelerate the solution for MCFP and make global optimization of routing between satellite network nodes. After the quantitative comparison between our MCFP algorithm and general augmented path maximum flow algorithm and exploring the detail of the algorithm, we verify the approximation algorithm’s reasonable selection of routing optimization in small satellite network node communication.

Space-Ground Integrated Network (SGIN) is the future network, and the satellite-earth link channel is one critical part of the SGIN. This paper simulates the satellite-earth link channel of SGIN based on the simulation environment of OMNeT++. We set up the model of space-ground network and satellite-earth link channel. The satellite-earth link channel includes two main parts, one part is the free space channel that ranges from the satellites to the aerosphere and the other part is the channel that ranges from aerosphere to the ground terminals. According to the ITU Recommendations, we simulate the satellite-earth link channel of the SGIN, from the results of the simulation. We analyze the satellite-earth link channel attenuation, obtaining the packet delay and packet arrival rate of the SGIN as well.

Automatic modulation classification (AMC) plays an important role in many fields to identify the modulation type of wireless signals. In this paper, we introduce deep learning to signal recognition. Based on architecture analysis of the convolutional neural network (CNN), we used real signal data generated by instruments as dataset, and proposed an improved CNN architecture to achieve compatible recognition accuracy of modulation classification. According to various conditions of signal noise ratio (SNR), we test the proposed CNN architecture with the real sampled signals. Experiments results show that the high-layer network is not necessary for modulation recognition with high SNR signals. The proposed CNN architecture has higher average classification accuracy than RESNET and is more compatible for modulation classification of signals with lower SNR.

With the difference of satellite altitude, there are always some inherent defects in the traditional single-layer satellite networks. In this paper, in order to improve the performance of the single-layer networks, a multi-layer satellite network model composed of LEO/MEO/GEO and inter satellite link is proposed. In this model, the LEO and MEO layers are used as the access layer, and the data transmission is carried out to the ground. As the core layer, the GEO layer is responsible for the management of the whole network and the link assignment. Then modeling the network based on the STK satellite simulation platform and carrying out the simulation analysis of ground coverage, the performance of the inter satellite link and the link transmission. Theoretical analysis and simulation results show that the design of multi-layer satellite network is reasonable and effective, and also can be used in the construction of the integrated satellite-terrestrial networks.

In order to improve the correct recognition rate of signals transmitted in satellite communication system, three different structures of artificial neural network (ANN), including feed forward network (FFN), cascade forward network (CFN) and competitive neural network (CNN) are investigated in this paper. Then their performance of correct recognition rate and performance of convergence rate are compared. Results of simulation indicate that typical FFN’s performance dramatically deteriorates in the case of Rician fading, CFN’s performance is similar to the former one while it has higher convergence rate. CNN’s performance of correct recognition rate is the best among these three nets, but in the training process, its performance of convergence rate is not good.

As one of the most important convergence layer (CL) protocol for delay/disruption-tolerant networking (DTN), Licklider transmission protocol (LTP) has recently been proposed for deep space communications, but it has rarely been considered for near earth applications. In this paper, LTP is adopted instead of TCP as CL with Bundle protocol (BP) for future application in GEO-relayed space networks (GRSN). Experiments are conducted on our computer based testbed in emulation of the basic scenarios during data transmission from LEO satellite to a ground station in GRSN. The results show that in transmission efficiency BP with LTPCL outperforms other protocols, such as BP with TCPCL, direct terrestrial TCP (TCP Cubic) and TCP variants (TCP Hybla) for space segments in most scenarios. It could be envisioned that DTN with LTPCL for space segment is currently the best choice for future GEO-relayed space internetworking.

This paper presents a design method of collimating optical system. LED has the characteristics of small size and long life. The performance of the optical system can be improved. A design of regular arrays is put forward in this paper. And this design can decrease the divergence angle through to the LED light source for the secondary light distribution. Besides the construction will be miniaturized and High-effectived.

Current stereo matching methods can be divided into 1D label algorithms and 3D label algorithms. 1D label algorithms are simple and fast, but they can’t aovid fronto-parallel bias. 3D label algorithms can solve fronto-parallel bias. However, they are very time-consuming. In order to avoid fronto-parallel bias efficiently, this paper introduces a new global depth refinement based on patches. The method transforms the depth optimization problem into a quadratic function computation, which has a low time complexity. Experiments on Motorcycle imagery and Wuhan university imagery verify the correctness and the effectiveness of the proposed method.

As one of the newest spatial information gathering methods, three-dimensional laser scanning technique is widely adopted in various fields due to its attributes of high accuracy and non-contact. However, currently, most systems of this kind are costly and with complex data post-processing requirements, which makes them not welcome enough for public usages. To deal with this, a novel terrain scanning system using line laser based on trigonometric survey is proposed. The system is capable of terrain data collection, data pre-processing, and 3D display. The data collection circuit is designed under Labview and PCL is applied for interface design. Collected data will be imported to the interface after pre-processing, thus providing the measured 3D terrain information. The experiment results show that the proposed system is capable of large area terrain scanning and display at a high speed and with low cost, and is more portable comparing to existing systems.

Contourlet transform lacks shift invariance, and threshold processing on the coefficients may produce pseudo Gibbs phenomena. For recursive cycle spinning algorithm can reduce the pseudo Gibbs phenomena. This paper studies the image denoising method combined with Contourlet transform and recursive cycles pinning algorithm, The analysis show that the factor need to be adjusted. When the adjustment factor takes best value, the corresponding image objective index PSNR (Peak Signal to Noise Ratio) is the largest, and images visual effects are optimal. The experimental results show that: compared with original algorithm, changing adjustment factor, the PSNR of denoised image can be improved 0.6–1.2.

In ultra-dense networks (UDNs), large-scale deployment of femtocells base stations is an important technique for improving the network throughput and quality of service (QoS). However, traditional resource allocation algorithms are concerned with the improvement of the overall performance of the network. In this paper, a new resource allocation algorithm based on game theory is proposed to manage the resource allocation in UDNs. The quality of service (QoS) and energy consumption of each femtocell are considered. Firstly, a modified clustering algorithm is performed. Then we transform this resource allocation problem to a Stackelberg game. In sub-channel resource allocation, we aim to maximize the throughput of the whole system by cluster heads (CHs). The power allocation takes account of the balance between QoS requirement and transmit power consumption. Simulation results show that this method has some advantages in improving the overall system throughput, while obtaining a performance improvement compared with other algorithms.

With the rapid growth of mobile data traffic demand, D2D relay technology is becoming an essential technology for the next generation mobile network. In order to select the optimal node in a shorter time, a cooperative D2D relay model considering the physical distance and social information is proposed. And then a threshold based on distance and social information is introduced, which is used to filter out the nodes with poor performance to get a relatively small candidate relay set. According to the optimal stopping theory, this paper presents a D2D relay optimal selection algorithm in order to weigh the consumption of exploration and system performance. The simulation results show that the algorithm proposed is superior to the traditional algorithm in system performance and algorithm complexity.

Large-scale multiple-input multiple-output (MIMO) system has the advantages of high energy efficiency and spectrum utilization. But using some cheap hardware may cause some problems, such as nonlinearity of the high power amplifier (HPA). When HPA works in the nonlinear region, it will affect the received signal and greatly reduce the performance of the system. In this paper, we first study the impact caused by nonlinear HPA, and then we optimize the traditional precoding algorithm to design an improved precoding algorithm which can reduce the impact. The simulation results show that the proposed algorithms perform better in bit error ratio and system capacity compared to the block of diagonalization (BD) precoding algorithm and forced zero (ZF) precoding algorithm, especially in the condition of high signal to noise ratio (SNR). So we can draw the conclusion that the algorithms proposed in this paper are able to reduce the impact caused by nonlinear HPA to the system.

The massive multiple-input multiple-output (MIMO) system is one of the most promising techniques, which extends degrees of freedom, increases the throughput of systems, supports more data streams and decreases transmit power. However, using cheap hardware in massive MIMO system can affect the overall performance of the system and deteriorate the user experience. The IQ imbalance caused by using cheap hardware is one of the important factors affecting system performance. To solve this problem, this paper proposes the design of precoding matrix based on the minimum mean square error (MMSE) criterion to suppress the influence of IQ imbalance on system performance. The numerical simulation results validate the effectiveness of the proposed algorithm, and show that the bit error rate (BER) performance of the proposed algorithm has obvious better than that of ZF, BD and WL-BD precoding.

With the development of IT technology, the traditional information technology cannot meet magnitude data analysis in GB level, let alone in TB level. So it is a perfect time for APACHE company to launch a new product, Hadoop framework, which is a JAVA based basic framework of distributed system, and the versions are now already designated as 2.X series, which means this Hadoop framework is one of the mainstream framework of massive data storage, data procession and analytical in this present.

The lens antenna array is typically composed of an electromagnetic (EM) lens and has elements in the focal area of the lens in order to achieve its large antenna gain. In this paper, we first analyze the response model of the lens antenna array, and conclude that the model follows the “sinc” function. The lens array is then applied to a MIMO system that allows millimeter-wave input and the use of new path-division multiplexing. On this basis, we model the channel of the system to derive the channel impulse response, which follows the “sinc sinc” function. Finally, the beamforming process is performed at the receiving end to obtain the received signal, and the signal-to-noise ratio expression is analyzed and optimized to obtain the maximum signal-to-noise ratio (SNR) of the system and the system performance is simulated.

In this paper, antenna selection (AS) is considered for clustering management (CM) to improve the spectrum efficiency of asymmetric interference networks. Through the proposed CM scheme, the whole network can be divided into several clusters, which will lead to a relative redundance of antenna resource for each interference alignment (IA) pair in the IA cluster. Therefore, the AS technique is adopted to improve the performance through selecting the optimal antenna combination for IA pairs. Considering the high computational complexity of the exhaustive search (ES) AS method, the cross-entropy optimization (CEO) algorithm is used to perform the IA technique, which can achieve relatively high performance with low computational complexity. From the simulation results, we can find that the proposed AS method in clustering management can further enhance the performance of the IA-based network.

Most of existing works on simultaneous wireless information and power transfer (SWIPT) for OFDM systems are studied based on power splitting or time splitting, which may lead to the time delay and the decreasing of sub-carrier utilization. In this paper, a multiuser orthogonal frequency division multiplexing (OFDM) system is proposed, which divides the sub-carriers into two parts, one for information decoding and the other one for energy harvesting. We investigate the optimization problem for maximizing the sum rate of users under the constraint of energy harvesting through optimizing the channel allocation and power allocation. By using the iterative algorithm, the optimal solution to the optimization problem can be achieved. The simulation results show that the proposed algorithm converges fast and outperforms the conventional algorithm.

With the development of integrated circuit, SoC systems are more and more used in products. Memory is an important part of SoC, SRAM design is a key research area. In this paper, based on ASIC design methodology, 2 K-bits SRAM is designed. A 6T-SRAM memory cell is designed and simulated with circuit level to improve reliability. The memory cell is used to construct the storage array, which are the word line 32 bits and the bit line 8 bits. Then, the SRAM peripheral circuit is designed and simulated by using SMIC 0.18 μm process, including the data input/output buffer circuit, clock circuit, address decoding circuit, data read/write circuit and sense amplifier. The structure, function and performance of latch type sense amplifier are analyzed emphatically. The simulation results demonstrate that the function of SRAM is verified correctly. The clock frequency of the SRAM can reach 100 MHz.

MANET has been widely used in many fields with the development of wireless communication technology. The AODV routing protocol which is known as a well-designed protocol of MANET has received widespread attention. However, high node velocity and frequent changes of network topology pose a challenge to the classic AODV protocol. Considering the stability of link, this paper proposes an algorithm to quantify the change frequency of network topology at first. Then a modified AODV protocol based on node velocity which is named RAODV is introduced in detail for high dynamic network topology. RAODV can build a more stable link according to the node velocity and reduce the normalized overhead of routing and average end-to-end delay by prolonging routing’s survival time.

Prime number generation and the large number operations directly affect the efficiency of RSA encryption algorithm. In order to reduce the number of the calculation process about modular operation and to reduce the difficulty of division in the calculation process, the Montgomery optimization algorithm is used to carry out the modular multiplication of RSA encryption algorithm, so that the efficiency of the algorithm is improved. Based on the application and research of hardware implementation to information encryption, the Verilog hardware description language is used to design the RSA encryption algorithm in 1024 bits. The simulation results of encryption and decryption experiment show that Montgomery modular multiplication algorithm and RSA encryption algorithm are verified to be correct and effective.

As the satellite networks can provide Internet access services, there are more and more kinds of data are transmitted on it. To ensure all kinds of data can be transmitted satisfied their own reliable requirements and obtain high transmission efficiency, a novel UEP transmission scheme based on distributed LT codes was proposed in this paper. In which scheme, the sub-codes on each node in the satellite network are performed with EEP property. By assigned different output degree distributions for the sub-codes, different kinds of data transmitted under the proposed scheme can be recovered by different reliable levels with nearly optimal transmission efficiency. On other hand, compared with the traditional distributed LT codes based transmission schemes, the relay nodes in proposed scheme do not have to know the reliable level of each source node, hence the security of the data can be guaranteed. We also make the asymptotic and finite-length analysis of proposed coding scheme, and the numerical results shows that the proposed scheme can provide UEP property between different kinds of data with low overhead performance, which can ensure the efficiency of data transmission.

A new class of rateless codes which are able to provide unequal error protection (UEP) and equal recovery time (ERT) properties is proposed in this paper. Existing UEP-based LT codes have an important property termed unequal recovery time (URT), which means the data with different reliability requirements can be recovered with different overhead, and it is worth noting that the most important bits (MIB) also have better recovery time performance. The proposed codes can recover data with the same overhead and different error performance. We analyze the asymptotic and experimental error performance of the proposed codes, and give the comparison between the proposed and traditional codes, our results show that the new class of UEP rateless codes are useful for scenarios in which the data have different reliability and same timeliness requirements.

In recent years, with the rapid development of wireless communication, the traditional cellular with isomorphic and regular structure has been unable to meet the increasing number of users and business needs involving data of big volume. The trend is evolving into Ultra Dense Network (UDN) architecture which is covered by cellular of irregular complex structure. In UDN, the spatial distribution of the base station plays an important role in the interference and performance evaluation of the whole cellular network, and the concept of green communication has also been put on agenda. In this paper, stochastic geometry theory is used to model UDN and to analyze the key performance of interference and wireless network. Moreover, a green communication strategy called TRSC is proposed, which is aimed at save energy and reduce the signal interference among cells to some extent.

Internet of Things (IoT) is going to create a world where physical objects are integrated into traditional networks in order to provide intelligent services for human-beings. Trust plays an important role in communications and interactions of objects in IoT. Two vital tasks of trust management are trust model design and reputation evaluation. However, current literature cannot be simply and directly applied to the IoT due to smart node hardware constraints, very limited computing and energy resources. Therefore a general and flexible model is needed to meet the special requirements for IoT. In this paper, we firstly design LTrust, a layered trust model for IoT. Then, a Reputation Evaluation Scheme for the Node (RES-N) has been presented. The proposed trust model and reputation evaluation scheme provide a general framework for the study of trust management for IoT. The efficiency of RES-N is validated by the simulation results.

Gain-phase error is inevitable in direction of arrival (DOA) estimation, it will lead to the mismatch between actual and ideal array manifold. Therefore, a novel gain-phase error calculation approach in DOA estimation for mixed wideband signals is provided in this paper. First, the signals are transformed on the focusing frequency. Then peak searching is employed for determining the far-field sources. Finally, gain-phase error can be calculated according to the orthogonality of far-field signal subspace and noise subspace, simulation results manifest the effectiveness of the proposed approach.

With the electromagnetic frequency getting higher and higher, the distance between the sensors is becoming smaller and smaller, so the mutual coupling is increasingly obvious, it will lead to the mismatch between actual and ideal array manifold. Therefore, a novel mutual coupling error calculation approach in direction of arrival (DOA) estimation for mixed wideband signals is provided in this paper. First, the signals are transformed on the focusing frequency. Then root finding is employed for determining the far-field signals. Finally, mutual coupling error can be calculated according to the orthogonality of far-field signal subspace and noise subspace.

Bandwidth and energy constraints of underwater wireless sensors networks necessitate an intelligent data transmission between sensor nodes and the fusion center. This paper considers a data gathering underwater networks for monitoring oceanic environmental elements (e.g. temperature, salinity) and only a portion of measurements from sensors allows for oceanic information map reconstruction under compressed sensing (CS) theory. By utilizing the spatial sparsity of active sensors’ data, we introduce an activity and data detection based on CS at the receiver side, which results in an efficient data communication by avoiding the necessity of conveying identity information. For an interleave division multiple access (IDMA) sporadic transmission, CS-CBC detection that combines the benefits from chip-by-chip (CBC) multi-user detection and CS detection is proposed. Further, by successively exploring the sparsity of sensor data in spatial and frequency domain, we propose a novel efficient data gathering scheme named Dual-domain compressed sensing (DCS). Simulation results validate the effectiveness of the proposed scheme and an optimal sensing probability problem related to minimum reconstruction error is explored.

The fact of multi-hop data transmission in wireless sensor network will lead serious load unbalance. Considering the limited energy supply, the load distribution will cause great restraints in relative applications. Existing algorithm mostly perform the load balance inside each cluster without considering about the entire network consumption. A cluster-based Balanced Energy Consumption Algorithm (BECA) is proposed by collecting the data more efficiently to avoid heavy traffic nodes so as to achieve global load balance. Simulating results show that BECA can obtain better balance properties and prolong the network lifetime greatly.

Electronic toll collection system (ETC) is usually used in Open road tolling (ORT) or free-flow tolling. In this project, ETC system is designed both in software and hareware. Radio frequency identification (RFID) technology is used to further improve performance of this system. This ETC system can be installed without tearing up the road, also this system can collect the data from the passing vehicles, identify the license plate and vehicle model and RFID electronic tag, retrieve the related registration informatiom from database, match registration information with the data ETC system collected, prevent switching license plates, collect the tolls automatically without having vehicles to slow down to pay.

This project is aiming to design a kind of survey robot that has the combined functions of the large-scale disaster search and rescue equipment and industrial surveillance camera, use virtual reality (VR) technology to improve human-robot interface, to provide more simpler way to present the true images of the survey environment. The whole design solution consists of three parts: survey vehicle, VR (virtual reality) display system, Hand grip remote control. Remote control can control survey vehicle mode conversion, robot movement, and high beam brightness adjustment. Data collected by survey vehicle are used to construct the image by VR imaging method, coupled with the VR on the camera point of the somatosensory remote control. This can enhance the sense of environmental immersion.

For the validation of eMMC device performance problems involving the effectiveness of testing and non-real time on parameters controlling, it may not be possible to obtain the performance data flexibly and efficiently, requiring consideration of the multi-channel parallel processing and real-time controlling. This paper presents a development platform for eMMC 5.0 device based on Zynq-7000. By combining hardware and software design, this platform is able to support eight eMMC devices working in parallel and get testing information in real time. Meanwhile, the device driver aims at achieving high performance data transfer by using DMA.

Under the background that SSD is more and more popular, this paper shows an efficient implementation of an SSD device designed for special function and interface on Xilinx SoC platform. The Device uses MLC NAND Flash as storage chips and uses Xilinx’s Zynq-7000 series SoC as the processor. The device adopts the method of multi-channel parallel data transmission and pipeline operation to achieve high performance. Enhanced ECC checking ability is provided to against the flash internal errors. The storage system also uses the RAID5 architecture to improve reliability significantly. Finally, the test results show that the designed SSD storage device reaches the expected performance and reliability.

Compared with the traditional Internet architecture, Delay/Disruption Tolerant Networking (DTN) has a bundle layer between the application layer and transport layer and is able to tolerate delays and disruptions. In this paper, we simulate typical DTN routing protocols and configure a vehicular network environment to evaluate the protocol performance. We compare the performance evaluation indicators that include delivery ratio, overhead ratio, average delay, and average buffer time with different node numbers and buffer sizes. We finish the simulation using Opportunistic Network Environment (ONE) simulator and the DTN routing protocols are evaluated according to the simulation results.

In sporadic machine-to-machine (M2M) communication, for the Code Division Multiple Access (CDMA) system with random access, applying compressed sensing (CS) algorithms to communication processes is a solution of multi-user detection (MUD). Many papers have shown that compressed sensing multi-user detection (CS-MUD) brings the benefits of jointly detecting activity and data. This paper focuses on the benefits of CS-MUD to the design of spread spectrum code in CDMA systems. Simulations show that CS-MUD brings two advantages in the spread spectrum code design: (1) There exist code sets with short code length can accommodate more users. (2) Code sets design is not limited to the design requirements of pseudo-random sequences, and the CS measurement matrix can be used as the code set. That is, CS-MUD provides a new idea for design and selection of spread spectrum code sets.

With the rapid development of wireless communication industry, the problem of spectrum resource scarcity is becoming more and more serious. To improve the spectral efficiency by compressing the adjacent carrier frequency intervals will increase the inter carrier interference. At this point, the performance of time-invariant filter is poor. If the communication system can make full use of the low energy slots in the time-frequency domain, the spectral efficiency could be improved. This paper proposes a time-frequency resource allocation method in which there is a delay of half a symbol between adjacent carriers. Accordingly, a time-varying filter is proposed. The simulation results show that the proposed time-varying filter performs better than the time-invariant filter.

The limitation of bandwidth, environment and multipath fading in wireless sensor network (WSN) cannot satisfy the need of users. Cooperative multiple-input-multiple-output (C-MIMO) technology is introduced to improve the communication performance, which brings in security problem as the same time. The key management technology may ensure the confidentiality with fewer keys but is unable to resist the compromised node attack. A new detection algorithm is proposed to sign the compromised node and recover the information during the transmission. Combining the key management and compromised node detection, a secure communication mechanism for WSN is proposed to resist the external and internal attack. Simulation results verify the advantages of security and performance by the proposed communication mechanism.

For the poor transmission reliability of data, this paper focuses on the precoding techniques of the multi-user MIMO system in the broadcast phase in, and the single-user MIMO and multi-user MIMO precoding techniques have been studied. In the multi-user MIMO system, BD-MMSE-VP coding algorithm is proposed to improve the BER performance of the system by using the MMSE criterion to optimize the perturbation vector, and the system uses QR technique to decompose BD channel matrix. Simulation results verify the effective of the proposed precoding algorithm.