Virtual Adversarial Training-Based Semisupervised Specific Emitter Identification

Deep learning is a new direction of research for specific emitter identification (SEI). Radio frequency (RF) fingerprints of the emitter signal are small and sensitive to noise. It is difficult to assign labels containing category information in noncooperative communication scenarios. This makes network models obtained by conventional supervised learning methods perform unsatisfactorily, leading to poor identification performance. To address this limitation, this paper proposes a semisupervised SEI algorithm based on bispectrum analysis and virtual adversarial training (VAT). Bispectrum analysis is performed on RF signals to enhance individual discriminability. A convolutional neural network (CNN) is used for RF fingerprint extraction. We used a small amount of labelled data to train the CNN in an adversarial manner to improve the antinoise performance of the network in a supervised model. Virtual adversarial samples were calculated for VAT, which made full use of labelled and large unlabelled training data to further improve the generalization capability of the network. Results of numerical experiments on a set of six universal software radio peripheral (USRP; model B210) devices demonstrated the stable and fast convergence performance of the proposed method, which exhibited approximately 90% classification accuracy at 10 dB. Finally, the classification performance of our method was verified using other evaluation metrics including receiver operating characteristic and precision-recall.

Dual-population Co-evolution Multi-objective Optimization Algorithm and Its Application

In the multi-objective optimization algorithm, the parameter strategy has a huge impact on the performance of the algorithm, and it is difficult to set a set of parameters with excellent distribution and convergence performance in the actual optimization process. Based on the MOEA/D algorithm framework, this paper construct an improved dual-population co-evolution MOEA/D algorithm by adopt the idea of dual-population co-evolution. The simulation test of the benchmark functions shows that the proposed dual-population co-evolution MOEA/D algorithm have significant improvements in IGD and HV indicators compare with three other comparison algorithms. Finally, the application of the LTE base station power allocation model also verifies the effectiveness of the proposed algorithm.

Uncertainty Quantification of Time-Dependent Quantities in a System with Adjustable Level of Smoothness

We summarise the results of a computational study involved with Uncertainty Quantification (UQ) in a benchmark turbulent burner flame simulation. UQ analysis of this simulation enables one to analyse the convergence performance of one of the most widely-used uncertainty propagation techniques, Polynomial Chaos Expansion (PCE) at varying levels of system smoothness. This is possible because in the burner flame simulations, the smoothness of the time-dependent temperature, which is the study's QoI is found to evolve with the flame development state. This analysis is deemed important as it is known that PCE cannot accurately surrogate non-smooth QoIs and thus perform convergent UQ. While this restriction is known and gets accounted for, there is no understanding whether there is a quantifiable scaling relationship between the PCE's convergence metrics and the level of QoI's smoothness. It is found that the level of QoI-smoothness can be quantified by its standard deviation allowing to observe the effect of QoI's level of smoothness on the PCE's convergence performance. It is found that for our flow scenario, there exists a power-law relationship between a comparative parameter, defined to measure the PCE's convergence performance relative to Monte Carlo sampling, and the QoI's standard deviation, which allows us to make a more weighted decision on the choice of the uncertainty propagation technique.

An Adaptive Reference Vector Adjustment Strategy and Improved Angle-Penalized Value Method for RVEA

Decomposition-based evolutionary multiobjective algorithms (MOEAs) divide a multiobjective problem into several subproblems by using a set of predefined uniformly distributed reference vectors and can achieve good overall performance especially in maintaining population diversity. However, they encounter huge difficulties in addressing problems with irregular Pareto fronts (PFs) since many reference vectors do not work during the searching process. To cope with this problem, this paper aims to improve an existing decomposition-based algorithm called reference vector-guided evolutionary algorithm (RVEA) by designing an adaptive reference vector adjustment strategy. By adding the strategy, the predefined reference vectors will be adjusted according to the distribution of promising solutions with good overall performance and the subspaces in which the PF lies may be further divided to contribute more to the searching process. Besides, the selection pressure with respect to convergence performance posed by RVEA is mainly from the length of normalized objective vectors and the metric is poor in evaluating the convergence performance of a solution with the increase of objective size. Motivated by that, an improved angle-penalized distance (APD) method is developed to better distinguish solutions with sound convergence performance in each subspace. To investigate the performance of the proposed algorithm, extensive experiments are conducted to compare it with 5 state-of-the-art decomposition-based algorithms on 3-, 5-, 8-, and 10-objective MaF1–MaF9. The results demonstrate that the proposed algorithm obtains the best overall performance.

Comparison Analysis on the Accuracy of Galileo PPP Using Different Frequency Combinations in Europe

The Galileo constellations are characterized by transmitting GNSS signals on multi-frequencies, which can benefit the robustness and accuracy of the solutions. However, the dual-frequency E1/E5a combinations are generally used for precise point positioning (PPP). In this paper, the performance of Galileo static and kinematic PPP using different dual- and multi-frequency combinations are assessed using observations from the European region. Overall, the accuracy of daily PPP achieved by the dual-frequency GPS, Galileo, and BDS is better than 5 mm in the horizontal direction and better than 10 mm in the vertical direction. Though the number of observed Galileo satellites is less than GPS, the horizontal accuracy can reach 1.6 mm/2.3 mm/5.7 mm on North/East/Up component, which is improved by 59.0% and 12.3% compared to the GPS in the north and up direction. Then, the accuracy of Galileo static PPP is analyzed using different dual- and multi-frequency combinations. Results indicate that the Galileo E1/E5b PPP can degrade the accuracy due to the inter-frequency clock biases between the E1/E5a and E1/E5b combinations. Best accuracy can be achieved for the triple- and four-frequency PPP, which is 4.8 mm in the up direction. The hourly accuracy for the static PPP can reach 5.6 mm/9.2 mm/12.6 mm in the north/east/up direction using the GPS/Galileo/GLONASS/BDS combinations. Finally, a positioning convergence ratio (PCR) indicator, which represents the accuracy of PPP over a period, is used to analyze the convergence time of kinematic PPP. Results indicated that the multi-frequency Galileo observations contribute minorly to the convergence of kinematic PPP. However, Galileo shows the best convergence performance for the single GNSS positioning, and the GPS/Galileo combined PPP achieved the best performance for the PPP using different GNSS combinations.

Noise control with Kalman filter for active headrest

A control strategy with Kalman filter (KF) is proposed for active noise control of virtual error signal for active headset. Comparing with the gradient based algorithm, KF algorithm has faster convergence speed and better convergence performance. In this paper, the state equation of the system is established on the basis of virtual error sensing, and only the weight coefficients of the control filter are considered in the state variables. In order to ensure the convergence performance of the algorithm, an online updating strategy of KF parameters is proposed. The fast-array method is also introduced into the algorithm to reduce the computation. The simulation results show that the present strategy can improve the convergence speed and effectively reduce the noise signal at the virtual error point.

Latency Aware Distributed ADMM over Networks

Methods based on the alternating direction method of multipliers (ADMM) has attracted academic attention because of its excellent convergence performance and potential application scenarios in many machine learning or optimization fields. However, classical distributed ADMM algorithm assumed ideal network communication, which do not consider the impact of network delay on computing performance. In this paper, based on the strategy of selecting bridges with lowest network latency and appropriate iterative process, we propose a latency aware distributed ADMM algorithm to alleviate the impact of network delay. The classical algorithm and proposed algorithm are tested and compared in real network scenarios. Experiments show that the proposed algorithm reduces the running time and improves the computing performance. Especially in networks with large delay, the effect is more obvious.

Parameter Optimization of dsRNA Splicing Evolutionary Algorithm Based Fixed-Time Obstacle-Avoidance Trajectory Planning for Space Robot

This paper addresses a smoother fixed-time obstacle-avoidance trajectory planning based on double-stranded ribonucleic acid (dsRNA) splicing evolutionary algorithm for a dual-arm free-floating space robot, the smoothness of large joint angular velocity is improved by 15.61% on average compared with the current trajectory planning strategy based on pose feedback, and the convergence performance is improved by 76.44% compared with the existing optimal trajectory planning strategy without pose feedback. Firstly, according to the idea of pose feedback, a novel trajectory planning strategy with low joint angular velocity input is proposed to make the pose errors of the end-effector and base converge asymptotically within fixed time. Secondly, a novel evolutionary algorithm based on the gene splicing idea of dsRNA virus is proposed to optimize the parameter of the fixed-time error response function and obstacle-avoidance algorithm, which can make joint angular velocity trajectory is planned smooth. In the end, the optimized trajectory planning strategy is applied into the dual-arm space robot system so that the robotic arm can smoothly, fast and accurately complete the tracking task. The proposed novel algorithm achieved 7.56–30.40% comprehensive performance improvement over the benchmark methods, experiment and simulation verify the effectiveness of the proposed method.

Geometric-algebra affine projection adaptive filter

Geometric algebra (GA) is an efficient tool to deal with hypercomplex processes due to its special data structure. In this article, we introduce the affine projection algorithm (APA) in the GA domain to provide fast convergence against hypercomplex colored signals. Following the principle of minimal disturbance and the orthogonal affine subspace theory, we formulate the criterion of designing the GA-APA as a constrained optimization problem, which can be solved by the method of Lagrange Multipliers. Then, the differentiation of the cost function is calculated using geometric calculus (the extension of GA to include differentiation) to get the update formula of the GA-APA. The stability of the algorithm is analyzed based on the mean-square deviation. To avoid ill-posed problems, the regularized GA-APA is also given in the following. The simulation results show that the proposed adaptive filters, in comparison with existing methods, achieve a better convergence performance under the condition of colored input signals.