Diffusion augmented complex adaptive IIR algorithm for training widely linear ARMA models

In this paper, our aim is to propose a fully distributed adaptive algorithm for learning the parameters of a widely linear autoregressive moving average model by measurements collected by a network. To this end, we consider a connected network where every node uses the augmented complex adaptive infinite impulse response (ACA-IIR) filter as the learning rule. We firstly formulate the learning problem as an optimization problem and resort to stochastic gradient optimization argument to solve it and derive the proposed algorithm, which will be referred to as diffusion ACAIIR (DACA-IIR) algorithm. We also introduce a reduced-complexity version of the DACA-IIR algorithm. We use both synthetic and real-world signals in our simulations where the results show that the proposed cooperative algorithm outperforms the noncooperative solution.     

On the utility of MIMO multi-relay networks for modulation identification over spatially-correlated channels

The necessity to perfectly monitor the intercepted signals for spatially-correlated multiple-input multiple-output (MIMO) systems, involves modulation identification algorithms. In this paper, we present an algorithm dedicated to the modulation identification for correlated MIMO relaying broadcast channels with direct link using multi-relay nodes. By modeling spatially-correlated MIMO channels as Kronecker-structured and the imperfect channel state information of both the source-to-destination and the relay-to-destination errors as independent complex Gaussian random variables, we firstly derive the ergodic capacity of the proposed transmission system. It turns out that the ergodic capacities improve with the number of relay nodes. Based on a pattern recognition approach using the higher order statistics features and the Bagging classifier, we show that the probability to distinguish among M-ary shift keying linear modulation types without any priori modulation information is enhanced compared to the decision tree (J48), the tree augmented naive Bayes, the naive Bayes using discretization and the multilayer perceptron classifiers. We also study the effect of increasing the number of relay nodes. Numerical simulations show that the proposed algorithm using the cooperation of multi-relay nodes with the source node can avoid the performance deterioration in modulation identification caused by both spatial correlation and imperfect CSI.     

一种复数赫布型学习算法及其在自适应ⅡR滤波器中的应用

本文提出了一种复数赫布类型的学习算法，用来训练一个输入、输出及权值均为复数的复神经网络。本算法的推导是基于TLS准则，而不是常用的LS或LMS准则。算法的收敛性及收敛结果通过将其应用到自适应复数ⅡR滤波器中来加以说明。     

基于等变化自适应源分离算法的滚动轴承故障信号自适应盲提取

针对复杂工况下滚动轴承故障信号盲提取问题,该文提出一种独立分量分析(ICA)中非线性函数自适应选择方法,解决了等变化自适应源分离算法(EASI)在多类振动源共存的情况下无法分离轴承故障信号的问题。此外,为了解决在线盲分离算法稳态误差与收敛速率的平衡问题,提出基于模糊逻辑的自适应迭代步长选择方法,极大地提高了学习算法的收敛速度,且稳态误差更小。轴承故障数据的盲提取仿真结果验证了算法的性能。     

Steady-state analysis of quantized distributed incremental LMS algorithm without Gaussian restriction

In this paper, we analyze the steady-state performance of the distributed incremental least mean-square (DILMS) algorithm when it is implemented in finite-precision arithmetic. Our analysis in this paper does not consider any distribution of input data. We first formulate the update equation for quantized DILMS algorithm, and then we use a spatial-temporal energy conservation argument to derive theoretical expressions that evaluate the steady-state performance of individual nodes in the network. We consider mean-square error, excess mean-square error, and mean-square deviation as the performance criteria. Simulation results are generated by using two types of signals, Gaussian and non-Gaussian distributed signals. As the simulation results show, there is a good match between the theory and simulation.     

Incremental Adaptive Strategies Over Distributed Networks

An adaptive distributed strategy is developed based on incremental techniques. The proposed scheme addresses the problem of linear estimation in a cooperative fashion, in which nodes equipped with local computing abilities derive local estimates and share them with their predefined neighbors. The resulting algorithm is distributed, cooperative, and able to respond in real time to changes in the environment. Each node is allowed to communicate with its immediate neighbor in order to exploit the spatial dimension while limiting the communications burden at the same time. A spatial-temporal energy conservation argument is used to evaluate the steady-state performance of the individual nodes across the entire network. Computer simulations illustrate the results.     

A CFAR adaptive subspace detector for second-order Gaussian signals

We study the problem of detecting subspace signals described by the Second-Order Gaussian (SOG) model in the presence of noise whose covariance structure and level are both unknown. Such a detection problem is often called Gauss-Gauss problem in that both the signal and the noise are assumed to have Gaussian distributions. We propose adaptive detectors for the SOG model signals based on a single observation and multiple observations. With a single observation, the detector can be derived in a manner similar to that of the generalized likelihood ratio test (GLRT), but the unknown covariance structure is replaced by sample covariance matrix based on training data. The proposed detectors are constant false alarm rate (CFAR) detectors. As a comparison, we also derive adaptive detectors for the First-Order Gaussian (FOG) model based on multiple observations under the same noise condition as for the SOG model. With a single observation, the seemingly ad hoc CFAR detector for the SOG model is a true GLRT in that it has the same form as the GLRT CFAR detector for the FOG model. We give an approximate closed form of the probability of detection and false alarm in this case. Furthermore, we study the proposed CFAR detectors and compute the performance curves.     

DOA estimation of noncircular signals for coprime linear array via locally reduced-dimensional Capon

We investigate the issue of direction of arrival (DOA) estimation of noncircular signals for coprime linear array (CLA). The noncircular property enhances the degree of freedom and improves angle estimation performance, but it leads to a more complex angle ambiguity problem. To eliminate ambiguity, we theoretically prove that the actual DOAs of noncircular signals can be uniquely estimated by finding the coincide results from the two decomposed subarrays based on the coprimeness. We propose a locally reduced-dimensional (RD) Capon algorithm for DOA estimation of noncircular signals for CLA. The RD processing is used in the proposed algorithm to avoid two dimensional (2D) spectral peak search, and coprimeness is employed to avoid the global spectral peak search. The proposed algorithm requires one-dimensional locally spectral peak search, and it has very low computational complexity. Furthermore, the proposed algorithm needs no prior knowledge of the number of sources. We also derive the Crámer-Rao bound of DOA estimation of noncircular signals in CLA. Numerical simulation results demonstrate the effectiveness and superiority of the algorithm.     

Scheduling with interference decoding: Complexity and algorithms

In this article we study the problem of scheduling wireless links in the physical interference model with interference decoding capability. We analyze two models with different decoding strategies that explore the fact that interfering signals should not be treated as random noise, but as well-structured signals. The first model makes use of successive interference cancelation, which allows the strongest signal to be iteratively decoded and subtracted from a collision, thus enabling the decoding of weaker simultaneous signals. The second model explores the fact that routers are able to forward the interfered signal of a pair of nodes that wish to exchange a message and these nodes are able to decode the collided messages by subtracting their own contribution from the interfered signal. We prove that the scheduling problem remains NP-complete in both models. Moreover, we propose a polynomial-time scheduling algorithm that uses successive interference cancelation to compute short schedules for network topologies formed by nodes arbitrarily distributed in the Euclidean plane. We prove that the proposed algorithm is correct in the physical interference model and provide simulation results demonstrating the performance of the algorithm in different network topologies. We compare the results to solutions without successive interference cancelation and observe that considerable throughput gains are obtained in certain scenarios.     

基于深度信念网络资源需求预测的虚拟网络功能动态迁移算法

针对5G网络场景下缺乏对资源需求的有效预测而导致的虚拟网络功能(VNF)实时性迁移问题,该文提出一种基于深度信念网络资源需求预测的VNF动态迁移算法。该算法首先建立综合带宽开销和迁移代价的系统总开销模型,然后设计基于在线学习的深度信念网络预测算法预测未来时刻的资源需求情况,在此基础上采用自适应学习率并引入多任务学习模式优化预测模型,最后根据预测结果以及对网络拓扑和资源的感知,以尽可能地减少系统开销为目标,通过基于择优选择的贪婪算法将VNF迁移到满足资源阈值约束的底层节点上,并提出基于禁忌搜索的迁移机制进一步优化迁移策略。仿真表明,该预测模型能够获得很好的预测效果,自适应学习率加快了训练网络的收敛速度,与迁移算法结合在一起的方式有效地降低了迁移过程中的系统开销和服务级别协议(SLA)违例次数,提高了网络服务的性能。     

An optimum step-size assignment for incremental LMS adaptive networks based on average convergence rate constraint

This paper presents an optimum step-size assignment for incremental least-mean square adaptive networks in order to improve its robustness against the spatial variation of observation noise statistics over the network. We show that when the quality of measurement information (in terms of observation noise variances) is available, we can exploit it to adjust the step-size parameter in an adaptive network to obtain better performance. We formulate the optimum step-size assignment as a constrained optimization problem and then solve it via the Lagrange multipliers approach. The derived optimum step-size for each node requires information from other nodes, thus with some justifiable assumptions, near-optimum solutions are derived that depend only on local information. We show that the incremental LMS adaptive network with near-optimal step sizes has improved robustness and steady-state performance. Simulation results are also presented to support the theoretical results.     

An Adaptive Learning Rate with Limited Error Signals for Training of Multilayer Perceptrons

Although an n‐th order cross‐entropy (nCE) error function resolves the incorrect saturation problem of conventional error backpropagation (EBP) algorithm, performance of multilayer perceptrons (MLPs) trained using the nCE function depends heavily on the order of nCE. In this paper, we propose an adaptive learning rate to markedly reduce the sensitivity of MLP performance to the order of nCE. Additionally, we propose to limit error signal values at output nodes for stable learning with the adaptive learning rate. Through simulations of handwritten digit recognition and isolated‐word recognition tasks, it was verified that the proposed method successfully reduced the performance dependency of MLPs on the nCE order while maintaining advantages of the nCE function.     

Globally convergent blind source separation based on a multiuser kurtosis maximization criterion

We consider the problem of recovering blindly (i.e., without the use of training sequences) a number of independent and identically distributed source (user) signals that are transmitted simultaneously through a linear instantaneous mixing channel. The received signals are, hence, corrupted by interuser interference (IUI), and we can model them as the outputs of a linear multiple-input-multiple-output (MIMO) memoryless system. Assuming the transmitted signals to be mutually independent, i.i.d., and to share the same non-Gaussian distribution, a set of necessary and sufficient conditions for the perfect blind recovery (up to scalar phase ambiguities) of all the signals exists and involves the kurtosis as well as the covariance of the output signals. We focus on a straightforward blind constrained criterion stemming from these conditions. From this criterion, we derive an adaptive algorithm for blind source separation, which we call the multiuser kurtosis (MUK) algorithm. At each iteration, the algorithm combines a stochastic gradient update and a Gram-Schmidt orthogonalization procedure in order to satisfy the criterion's whiteness constraints. A performance analysis of its stationary points reveals that the MUK algorithm is free of any stable undesired local stationary points for any number of sources; hence, it is globally convergent to a setting that recovers them all.     

采用合成方法的多路信号自适应时延联合估计

本文关注的是多路信号之间时延差异的联合估计问题。不同于传统的自适应时延估计算法,本文以合成信号作为自适应时延估计的参考信号,给出了基于信号合成的联合自适应时延估计算法。同时本文推导和仿真了该算法时延估计的均值、学习曲线及方差特性。性能分析和仿真结果均显示,本文提出的基于合成的多路信号自适应时延估计为渐进无偏的时延估计。在不明显增加计算量的条件下,当算法收敛时,联合时延估计算法的方差显著低于传统的两路信号之间自适应时延估计算法方差。      

分布式子带自适应滤波算法

为了解决分布式最小均方算法在输入信号相关性较高时收敛速度较慢、分布式仿射投影算法计算复杂度较高等问题,本文提出了两种分布式子带自适应滤波算法,即递增式和扩散式子带自适应滤波算法.分布式子带自适应滤波算法将节点信号进行子带分割来降低信号的相关性,从而加快收敛速度.由于用于子带分割的滤波器组中包含了抽取单元,所以分布式子带自适应滤波算法和对应的分布式最小均方算法的计算复杂度相近.仿真结果表明,与分布式最小均方算法相比,分布式子带自适应滤波算法具有更好的收敛性能.     

An Efficient Adaptive Channel Estimation Algorithm for MIMO OFDM Systems—Study of Doppler Spread Tolerance

In this paper we develop an adaptive MIMO channel estimation algorithm for space–time block coded OFDM systems. The presented algorithm is based on Expectation Maximization (EM) technique by decomposing the superimposed received signals into their signal components, and estimating the channel parameters of each signal component separately. We also study and compare our proposed EM-based algorithm with a previously introduced recursive-least-squares based algorithm for MIMO OFDM systems. At each iteration the EM algorithm decomposes the problem of multi-channel estimation into channel estimation for each transmit–receive link. In this paper we also study the Doppler spread tolerance of our proposed algorithm in a fast fading environment, and investigate how it affects the system BER performance.     

Structural sparse coding seeds–active appearance model for object tracking

In this paper, we propose a tracking algorithm that can robustly handle appearance variations in tracking process. Our method is based on seeds–active appearance model, which is composed by structural sparse coding. In order to compensate for illumination changes, heavy occlusion and appearance self-updating problem, we proposed a mixture online learning scheme for modeling the target object appearance model. The proposed object tracking scheme involves three stages: training, detection and tracking. In the training stage, an incremental SVM model that directly measures the candidates samples and target difference. The proposed mixture generate–discriminative method can well separate two highly correlated positive candidates images. In the detection stage, the trained weighted vector is used to separate the target object in positive candidates images with respect to the seeds images. In the tracking stage, we employ the particle filter to track the object through an appearance adaptive updating algorithm with seeds–active constrained sparse representation. Based on a set of comprehensive experiments, our algorithm has demonstrated better performance than alternatives reported in the current literature.     

Fairness constrained diffusion adaptive power control for dense small cell network

Small cell is an emerging and promising technology for improving hotspots coverage and capacity, which tends to be densely deployed in populated areas. However, in a dense small cell network, the performances of users differ vastly due to the random deployments and the interferences. To guarantee fair performance among users in different cells, we propose a new distributed strategy for fairness constrained power control, referred to as the diffusion adaptive power control (DAPC). DAPC achieves overall network fairness in a distributed manner, in which each base station optimizes a local fairness with little information exchanged with neighboring cells. We study several adaptive algorithms to implement the proposed DAPC strategy. To improve the efficiency of the standard least mean square algorithm (LMS), we derive an adaptive step-size logarithm LMS algorithm, and discuss its convergence properties. Simulation results confirm the efficiency of the proposed methods.     

一类基于非线性PCA准则的复数信号盲分离算法

在阵列信号处理过程中，经常遇到复数信号盲分离问题。例如，卷积混合型的源信号的盲分离；声纳信号盲分离。本文提出了一类基于非线性准则的复数信号盲分离算法。将非线性函数引入学习过程，由算法自动调节学习速率。计算机仿真实验验证了算法的有效性，文中给出了验证结果。     

复高斯混合模型分布式语音分离方法研究

本文研究空域协方差矩阵初始化对复高斯混合模型下的分布式语音分离性能的影响.在不同节点的接收信号向量条件独立性假设前提下,推导出一种逐节点迭代更新所有接收信号向量对应的空域协方差矩阵和后验概率等参数的方法;基于此,本文提出用基于到达角度的导向矢量的相关矩阵初始化每个节点对应的空域协方差矩阵;同时,为保证不同节点能协同工作...