Abstract:In internet of things, the design of multiuser access system will meet great challenges. In this paper, a two-dimensional fountain system of multiuser access and SLT coding is presented based on the existing system. First of all, each user's original information is encoded by SLT codes on the time layer, and then each user competes for transmission according to the access probability on the user layer to form a two-dimensional fountain. The system makes full use of fountain codes to effectively improve the signal anti-interference ability. Secondly, for each user using SLT codes, we can simply employ the belief propagation decoding algorithm to decode. Simulation results show that compared with the existing multiuser access system, the proposed system can significantly achieve better throughput performance and solve the problem that the performance of the existing system decoding algorithm has a sharp decline in the higher access probability, so then can provide more users with the opportunity of information transmission which has good practicability in multiuser data transmission.

Abstract:This paper generalizes the traditional electromagnetic vector uniform array as an electromagnetic vector coprime array, and breaks the limit that the spacing between array elements is not more than half a wavelength. A direction of arrival (DOA) and polarization estimation algorithm is proposed for electromagnetic vector sensor coprime array via reduced-dimension Capon. The proposed algorithm does not need to assume that the polarization information is known, and only one-dimensional search is needed to avoid multi-dimensional search and enables automatic matching of DOA and polarization parameters. Compared with a uniform array with the same array element number, it significantly improves the angle estimates performance, and expands the antenna aperture. It also has a relatively high degree of freedom and lower complexity. Under the condition of same array and parameters, the angle estimation performance of the proposed algorithm is better than that of the ESPRIT algorithm and the trilinear decomposition algorithm.

Abstract:Extreme learning machine (ELM) is a randomized algorithm which randomly generates the input weights and hidden nodes biases of single-hidden layer feed-forward neural networks (SLFNNs), and then determines the output weights analytically. Given the architecture of SLFNN, we can obtain different learning models by repeatedly training SLFNNs with ELM. The paper proposes an approach by integrating these learning models for data classification. Specifically, firstly several SLFNNs are trained by ELM. Secondly the trained SLFNNs are integrated by majority voting method. Finally the integrated model is used for data classification. We experimentally compared the proposed approach with ELM and ensemble ELM (EELM) on 10 data sets. Experimental results show that the proposed approach outperforms ELM and EELM.

Abstract:A new approach is proposed to improve BPSK signal parameter estimation performance under low SNR environment. First, the smooth pseudo Wigner-Ville distribution(SPWVD) is applied to the BPSK signal and time-frequency matrix S_SPWVD (t_m,f_n) is obtained, which can estimate carrier frequency of the signal combined with the minimum entropy criterion. What's more, the discontinuity of signal phase is mapped to negative magnitude peaks on its carrier frequency in the time-frequency domain. Then,the code width of the BPSK signal can be estimated by searching the interval between adjacent negative peaks, and then the code rate, code number and phase-coded sequence could also be estimated. Finally, simulations show the validity of the proposed approach.

Abstract:To overcome the shortcoming that the single-state channel modeling method cannot sufficiently reflect the practical propagation of satellite channel, a three-state Markov modeling method for land mobile satellite channel is presented. In particular, it is the first time to simulate the symbol error rate (SER) performance based on the presented channel model. Each state of the three-state Markov channel model obeys the Loo distribution with different parameters; while the state transition is modeled as Markov random process. The SER comparisons using the presented channel model are simulated independently for the intermediate tree shadowed scenario and suburban scenario in S band, elevation of 40° and quadrature phase shift keying(QPSK) modulation. The simulation results show that SER performance is mainly determined by the strength of the line-of-sight component. In case of heavy shadow, additional signal to noise ratio (SNR) of 3-5 dB is needed to compensate the SER loss. In addition, in order to achieve the same SER=0.01 performance, the intermediate tree shadowed scenario needs extra about 5 dB SNR compared to suburban scenario. The work is contributed to the research of satellite communication techniques and the engineering application therein.

Abstract:In the paper, an iterative group sparse channel estimation and decoding algorithm is proposed for OFDM communication systems. The group basis pursuit de-noising (Group-BPDN) method based on the group sparsity of wireless multipath channel is used in channel estimation to improve the performance. Moreover, the soft-output viterbi algorithm(SOVA) method is used for channel decoding. The most reliable decoding data are fed back to the input of channel estimator. Consequently, the new set of known signals is constituted by the feedback data and pilot. Therefore, during iterative process, the known information used in the sparse recovery is increased, which can improve the performance of sparse recovery. Furthermore, the noise power which is useful for sparse recovery is estimated through the new set of known signals and the channel response is estimated. The simulation results demonstrate that the proposed algorithm improves the channel estimation performance and reduces the error rate.

Abstract:LDPC code recognition is a difficult issue in channel coding recognition. With the widespread application of LDPC code in communication field, LDPC code recognition technology has been attracted more and more attention. Aiming at the problem that the existing methods have been suffering weak identification performance in low SNR environment, a LDPC coding method with maximum mean square ratio is proposed in this paper. Firstly, the coded verification relation is mapped to the log-likelihood ratio domain using the soft information output in the channel, and the parity check log-likelihood ratio (CLLR) is defined. Then, the modulus statistical properties of CLLR are analyzed, and the relationship between CLLR and the parameters of LDPC code is established. Finally, using the difference of statistical properties of CLLR under different check matriices, a method of maximum mean and variance ratio based decision device is proposed. The simulation results show that the proposed algorithm is superior to the existing algorithms in the finite set application model, and the recognition gain can reach 2 dB to 5 dB in the low SNR environment. Moreover, for the recognition of high bit rate LDPC codes, the algorithm can significantly improve the recognition performance.

Abstract:In cognitive radio network environment, the base station needs to carry out an effective spectrum management policy to guarantee the licensed user's communication and to improve the quality of service of the cognitive radio users at the same time. In the process of allocating spectrum holes to cognitive radio users, the base station faces massive passive channel switching due to the unpredictability of the licensed user and it results in the throughput of cognitive radio users' degradation. To solve this problem, this paper proposes a novel base station-cognitive base station, which contains reinforcement learning model with novel state and action sets. The cognitive base station can perform two-step decision of channel allocation, that is, whether to switch the channel for cognitive radio users and how to select the best channel if the cognitive base station decides to switch, so as to avoid excessive channel switching and improve the throughput of the cognitive radio user. Also, the performance of reinforcement learning spectrum management policy highly depends on the exploration of environment. In this paper, epsilon-greedy exploration method is used to solve the balance problem of cognitive base station in exploring the unknown environment and exploiting the existing knowledge. Simulation results show that the implementation of the epsilon-greedy in each decision step has a remarkable effect on the system performance. Also, we set up the best evaluation of a combination of two-step epsilon so that the proposed method is superior to traditional reinforcement learning spectrum allocation scheme in improving cognitive radio users' throughput and reducing channel switching.

Abstract:In order to address the problem of spectrum sensing with random presence and absence of primary user's signal in cognitive radios, a novel energy based spectrum sensing method is proposed. The method first assumes that the departuring process and the arriving process of the primary user's signal are possion random processes. When a cognitive user tries to access a spectrum, the energy of the collected samples is combined linearly according to the departure rate of Possion random process and then the corresponding threshold is calculated to decide whether there is a primary user or not. If the primary user is absent, the cognitive user accesses and the spectrum is used. When the cognitive user is using the spectrum, it also combines the energy of the collected samples linearly according to the arrival rate of Possion random process and calculates the corresponding threshold to test whether the primary user is arriving or not. If the primary user is arriving, the cognitive user vacates this spectrum immediately. Simulation results show that compared with traditional energy based detection method, the proposed method can work efficiently with random presence and absence of primary user's signal.

Abstract:As a typical big data, data stream has the features of continuous, infinite, concept drift and fast arrived. The features make it impossible to apply traditional classification techniques to classify data streams. The paper proposes the concept very fast decision tree(CVFDT) update ensemble(CUE) algorithm based on the classic accuracy weighted ensemble (AWE) algorithm. This algorithm not only improves the weight distribution of the base classifier, but also improves the sensitivity of the block size and the increase of the dissimilarity between base classifiers. Experiments show that, in the classification accuracy, CUE algorithm is higher than the AWE algorithm. Finally, the dynamic classifier selection with clustering (DCSC) algorithm is proposed, which is based on the idea of classifier dynamic selection. The time efficiency is relatively high because there is no tedious weight value mechanism. Experimental results show that the DCSC algorithm can effectively handle the concept of drift and its efficiency is relatively high.

Abstract:As the demand of user's communication rate and data service is gradually increasing, the communication rate and spectrum resource of cellular network cannot meet the business requirements correspondingly. The equipment of direct communication (Device-to-device, D2D) can improve the coefficient of spectrum utilization and the overall communication capacity by reusing the spectrum resources of cellular users. However, reusing cellular network spectrum resources can cause serious interference and affect the overall quality of communication. In order to mitigate interference, the linear programming problem and the corresponding optimal algorithm of resource allocation are studied by establishing the communication system model. Considering the high complexity of the optimal algorithm, a heuristic algorithm to allocate the communication resource is proposed. By traversing the interference matrix between the D2D users and the cellular users to get the minimum value, the multiplexing resources are assigned to the corresponding D2D user and cellular user. After the cellular users are allocated to the communication resources, the D2D users are assigned the dedicated communication resources. Simulation results show that this algorithm can significantly reduce the interference of the D2D users to the cellular users and increase the number of D2D users to the maximum extent.

Abstract:The identification of interleaver type is the precondition for the identification of interleaver parameters in non-cooperative communication. A blind identification algorithm of interleaver type based on normalized rank characteristics for RS code is proposed. Unlike the prior works, the interleaver type identification process is carried out for a generic case without any restriction on the codeword length n. Combining the ratio of number of zeros in the column with Gauss-Jordan elimination through pivoting algorithm, the normalized rank can be determined in the presence of bit errors. Finally, the blind identification can be realized through the differences of the normalized rank between block interleaver and convolutional interleaver. Experimental results show that this proposed method can achieve a better performance compared with the existing algorithms in high bit error rate and can realize the identification of interleaver type. It has a certain guiding significance for practical engineering applications.

Abstract:Aiming at the problem that the data traffic will be delayed at relays and the delay requirement of it may not be satisfied, a relay selection model with the consideration of the relaying delay is established in the two hop DF cooperative communication systems. On this basis, a new relay selection method considering the relaying delay of the service is proposed. The performances of the proposed method are also analyzed. First of all, the probability density function (PDF) of the relaying delay at the relays is derived. Thereafter, the analytical expressions of the average system capacity and outage probability are obtained. The simulation results are in accordance with the theory accurately and also verify that the proposed method can effectively reduce the average delay at relays while guaranteeing a considerable data rate.

Abstract:Since the unmanned aerial vehicle (UAV) has high maneuverability and simple deployment, the UAV-based transmission technology has received much attention. As an important resource of the communication system, power allocation will affect the each link performance and the energy efficiency. In this paper, taking the energy efficiency as a criterion, a power allocation algorithm is proposed for UAV relay communication system under rician fading channel. First, the problem of power allocation is formulated as an optimization model for amplify-and-forward (AF) relay transmission model. Then the beamforming optimization scheme is obtained with the fixed transmission power. Next, the original non-convex optimization problem is transformed into convex optimization problem through large signal to noise ratio interval approximation. Finally, Karush-Kuhn-Tucker (KKT) condition is used to calculate the closed form of the optimal power allocation. Simulation results show that the performance of proposed algorithm is very close to that of the iterative method. But the computational complexity is lower. Compared with the average algorithm, the proposed algorithm has more larger energy efficiency which can lead to system performance improvement.

Abstract:Data segmentation is one of critical issues of model selection of parallel/distributed machine learning, which has impacts on generalization performance and parallel efficiency of parallel/distributed machine learning. Existing approaches to data segmentation of parallel/distributed machine learning are dependent on empirical evidences or on the number of the processors without explicit criterion. In this paper, we propose a parallel efficiency sensitive criterion of data segmentation with generalization theory guarantee, which improves the computational efficiency of parallel/distributed machine learning while retaining test accuracy. We first derive a generalization error upper bound with respect to the block number of the data segmentation. Then we present a data segmentation criterion that is a trade-off between the generalization error and the parallel efficiency. Finally, we implement large-scale Gaussian kernel support vector machines (SVMs) in the random Fourier feature space with the alternating direction method of multipliers (ADMM) framework on high-performance computing clusters, which adopt the proposed data segmentation criterion. Experimental results on several large-scale benchmark datasets show that the proposed data segmentation criterion is effective and efficient for the large-scale SVMs.

Abstract:Traditional clustering algorithms can not meet the requirements of current big data processing because of the limitations of stand-alone memory and computing power. Therefore it is urgent to find new solutions. Aiming at problems occurred in stand-alone memory calculating, combined with iterative computing features of clustering algorithms, a clustering system based on Spark platform is proposed. For the two different types of data sets, which are sparse sets and dense sets, the system firstly uses different strategies to achieve data preprocessing. Secondly, the performance of different clustering algorithms on Spark platform is analyzed and the best solution is given. Finally, the computing speed is improved with data persistence technology. Experimental results show that the proposed system can effectively meet the requirements of massive data clustering analysis.

Abstract:With the rapid development of radio technology and the rapid change of radio cheating means, the use of radio equipment in the examination of cheating has also increased. In order to ensure the fairness of the examination, how to find and locate the radio cheating signal effectively and construct the intelligent examination room has become a new hot topic that needs to be solved urgently. In view of this, on the basis of indoor location and spectrum monitoring technology, a discovery and location system of radio cheating signal based on deep learning is designed in this paper, which realizes the functions of judgment, location, alarm for the radio cheating signal, and real-time displaying on the mobile terminals. The system provides an intuitive, remote and real-time examination environment for invigilators, and creates a fair competition environment for the examinees.

Abstract:This paper presents an ultrasonic indoor positioning system which is easy to be deployed. Ultrasonic signal whose bandwidth is 20-22 kHz can be transmitted by ordinary tweeters. This paper realizes the time difference of arrival (TDOA) positioning based on ultrasound transmitted by three tweeters. A mobile phone collects ultrasonic signal in this band by microphone. A high accuracy, low-cost indoor positioning can be achieved after signal processing. Signal structure of ultrasonic positioning acoustic source is designed at the transmitter. A synchronization method based on frequency search and FFT search is used at the receiver. A modified SR-1 method is used in positioning scenario based on TDOA. This paper validates that the accuracy of modified SR-1 method is basically the same as the classical CHAN algorithm when three tweeters are used in the scenario. The band of the ultrasonic signal which is used in our scheme is fairly different from those commonly-used scheme whose signal band is 40 kHz. The error is less than 9 cm in every experiment under the condition of receiving ultrasonic signal by an ordinary phone.

Abstract:A discrete density evolution algorithm for Turbo decoding message passing(TDMP) decoding algorithm is proposed to solve the problem of fixed parameter selection in efficient LDPC decoder design. By using the discrete density evolution algorithm, the modification factors and the quantization precision in the decoding algorithm are optimized. Compared with the traditional method, the efficiency is greatly improved and the effect is significant. Experimental results show that the performance of the optimized fixed-point decoder is only about 0.1 dB worse compared with the pure floating-point simulation. In the structure design of the decoder, a P-message circular memory structure based on distributed RAM is proposed. Compared with the traditional memory structure based on register and Benes network, the resource consumption is obviously decreased. The hardware implementation and test on FPGA platform of Xilinx company show that it has some advantages in terms of resource consumption and throughput compared with the same kind of decoder, and it is an efficient LDPC hardware decoder.

Abstract:The temperature detection in the electrical power system is a principal method to avoid the equipment failure or accidents. In recent years, the use of the wireless temperature monitoring technique is an important manifestation of the smart electrical power system. In view of the development demand of smart grid, this paper designs a temperature detection node which is based on radio frequency communication technology. The node is composed of temperature sensor DS18B20, ultra-low power SCM PIC18LF14K50 and radio frequency transceiver chip MRF49XA, which can realize three working conditions of dormant wait, data sampling and wireless communication. By setting up the key parameters under different working conditions, the average power consumption of the temperature detection nodes is tested and compared, and the power consumption rules of the nodes are summarized. On this basis, a self-powered supply management unit based on LTC3588-1 chip is designed for the temperature detection node, which adopts the mature and reliable inductive current energy take-up technology to meet the self-powered supply requirements of the temperature detection nodes. The research shows that, when the node is in full electric state, once the energy source is cut off, the node can continue to work steadily for 47 h, which can maintain the normal work of the temperature detection node until the power grid is restored to normal operation.