In this paper, we proposed a generative model for feature selection under the unsupervised learning context. The model assumes that data are independently and identically sampled from a finite mixture of Student's distributions, which can reduce the sensitiveness to outliers. Latent random variables that represent the features' salience are included in the model for the indication of the relevance of features. As a result, the model is expected to simultaneously realize clustering, feature selection, and outlier detection. Inference is carried out by a tree-structured variational Bayes algorithm. Full Bayesian treatment is adopted in the model to realize automatic model selection. Controlled experimental studies showed that the developed model is capable of modeling the data set with outliers accurately. Furthermore, experiment results showed that the developed algorithm compares favorably against existing unsupervised probability model-based Bayesian feature selection algorithms on artificial and real data sets. Moreover, the application of the developed algorithm on real leukemia gene expression data indicated that it is able to identify the discriminating genes successfully.

This paper presents a novel backoff algorithm in CSMA/CA-based Medium Access Control (MAC) protocols for clustered sensor networks. We first show that every node should have the same value of Contention Window (CW) in a cluster by formulating resource allocation as a utility maximization optimal problem, then assume all nodes have the same CW and gain the relation between the optimal value of CW and the number of nodes by maximizing the total network utility with constrains of minimizing collision probability. The result is a new retransmission algorithm that uses an optimal shared CW that is easy to implement and results in fewer collisions than binary exponential backoff algorithm. The proposed scheme can decrease delay and improve throughput, moreover, it is also energy-efficiency for clustered sensor networks, simulation results validate our conclusion.

Wireless sensor networks (WSNs) are energy- constrained in nature, moreover, sensor nodes play the dual role of data gathering and data relaying. In this paper, we consider how to allocate the power of sensor nodes for forwarding traffic of other nodes. After forwarding power ratios of nodes are decided, we consider pricing as a mean to stimulate cooperation between a node and other nodes along its routing path to a sink node. By formulating the problem of data sensing and transport in WSNs as a network utility maximization (NUM) problem, we propose an iterative price and power adaption algorithm by using dual decomposition techniques. Numerical results show that by using our price mechanism, we can improve system performance while reducing power consumption.

Wireless sensor networks (WSNs) are energy- constrained in nature, in this paper, we formulate the problem of data transport in sensor networks as a network utility maximization (NUM) problem, but we argue that each source utility not only depends on its source rate, but also on the consumed energy, this leads to a coupled utility model, where the utilities are functions of source rates and consumed energy. Differentiating from the classical NUM framework which usually takes the consumed energy as constraints. Our utility model regards consumed energy as one of the components of measure of the utility values, which indicates the tradeoff of source rates and consumed energy, it is a more accurate utility model for abstracting the energy characteristics for data gathering and transmission in WSNs. Due to the coupled energy utility, our optimization problem is not separable. Despite the difficulty, we present a systematic approach to decouple our NUM problem with coupled utilities by introducing into the slack variables and using dual decomposition techniques, and obtain a distributed algorithm for solving our problem. The proposed algorithm can converge to the Pareto optimal tradeoff between rates and energy for all users.

Wireless sensor networks (WSNs) are energy- constrained in nature, in this paper, we formulate the problem of data transport in sensor networks as a network utility maximization (NUM) problem, but we argue that each source utility not only depends on its source rate, but also on the consumed energy, this leads to a coupled utility model, where the utilities are functions of source rates and consumed energy. Differentiating from the classical NUM framework which usually takes the consumed energy as constraints. Our utility model regards consumed energy as one of the components of measure of the utility values, which indicates the tradeoff of source rates and consumed energy, it is a more accurate utility model for abstracting the energy characteristics for data gathering and transmission in WSNs. Due to the coupled energy utility, our optimization problem is not separable. Despite the difficulty, we present a systematic approach to decouple our NUM problem with coupled utilities by introducing into the slack variables and using dual decomposition techniques, and obtain a distributed algorithm for solving this problem. The proposed algorithm can converge to the Pareto optimal tradeoff between rates and energy for all users.