Xiaobei Wu

Multiclass Least Squares Auto-Correlation Wavelet Support Vector Machines

In this paper, combining the auto-correlation wavelet kernel with multiclass least squares support vector machine (MLS-SVM), a novel notion of multiclass least squares support vector machine with universal auto-correlation wavelet kernels (MLS- AWSVM) is proposed. The translation invariant property of the kernel function enhances the generalization ability of the LS-SVM method and the spiral multiclass classification experimental results show some advantages of MLS-AWSVM over MLS- SVM on the classification and the generalization performance.

Adaptive Topology Reconfiguration from an Actor Failure in Wireless Sensor-Actor Networks

Wireless Sensor-Actor Network (WSAN) usually consists of numerous sensor nodes and fewer actors, and the connectivity of interactors is critical to the whole network. Due to the hash deployed environments and limited energy supply, actor nodes may fail and impact the performance of the whole network. Since the failure of a cut-vertex will disrupt connectivity and divide the topology into disjoint segments, most of the previous researches have already considered this scenario. However, the impact of an abruptly actors failure to the network will be far more than that. This paper focuses on the problem of an actors failure and gives a more comprehensive view of the faulty actor, that is, not only restricts to the cut-vertex. Length-Aware Topology Reconfiguration Algorithm (LTRA) is proposed on the basis of two vital definitions named as length impact index (LII) and vertex cut set (VCS). LTRA is a hybrid method which selects a best candidate for each actor (if it has) and then initiates in a distributed manner. Main idea of this approach is that candidate will move to replace the faulty one once the failure occurs. In addition, the candidate is selected from one-hop neighbors of each actor. Finally, performance of LTRA is validated by extensive simulation experiments.

Multiclass Least Squares Wavelet Support Vector Machines

A novel admissible support vector (SV) kernel, namely modified L-P wavelet kernel, is proposed based on theoretic analysis. The wavelet kernel can approximate arbitrary curve in quadratic continuous integral space, thus the generalization ability of the support vector machines (SVM) is improved. Based on the wavelet kernel function and the multiclass least squares support vector machines (MLS-SVM), the multiclass least squares wavelet support vector machines (MLS-MLPWSVM) is presented. The spiral multiclass classification experimental results show some advantages of MLS-MLPWSVM over MLS-SVM on the classification and the generalization performance.

Fault detection for networked control system with time-delay and data packet dropout

Focusing on the networked control system with network-induced time-delay and transmitted data packet dropout, based on assumptions that the rate of the data packet dropout is fixed and the network-induced time-delay is less than one sampling period, problems of observer-based fault detection of the system are studied. According to conditions of data arrival of the controller, the state observers of the system are designed to detect faults when they occur in the system. When the system is normal, the observer system is modeled as an asynchronous dynamic system with rate constraints on events. Based on the model, stability of the whole system is given. When conditions are satisfied, the system is exponentially stable. When a fault occurs, the observer residual can change rapidly to detect the fault. A numerical example shows the effectiveness of the proposed method.

Monitoring model of Wireless sensor networks based on state

Network model is the foundation of network performance optimization and network practical applications. Some basic network monitoring requirements were presented by analyzing the issues associated with existing models and node deployment of Wireless sensor networks (WSNs). Focusing on the network monitoring performance, a monitoring model of state based WSNs was proposed. The state information of single node was denoted as a vector in the model, including work state, position state, energy state, sensing and communication relations of nodes. A WSN could be denoted as a matrix which was composed of state vectors of nodes. Several examples were illustrated to demonstrate the applicability of the proposed model in some researches of WSNs, including energy control, topology reconfiguration, and performance evaluation. The model provides a platform to evaluate capabilities, functionalities, and behaviors of a WSN from the mathematics perspective.

Least-squares auto-correlation wavelet kernel method for regression estimation

Based on the auto-correlation wavelet kernel, a novel notion of least squares support vector machine (LS-SVM) with universal auto-correlation wavelet kernels is proposed for approximating arbitrary nonlinear functions. The translation invariant property of the auto-correlation wavelet kernel function enhances the generalization ability of the LS-SVM method and some experimental results are included to demonstrate the efficiency and validity of the proposed method.