Wilfried N. Gansterer

Static vs. mobile sink

Over the last decade a large number of routing protocols has been designed for achieving energy efficiency in data collecting wireless sensor networks. The drawbacks of using a static sink are well known. It has been argued in the literature that a mobile sink may improve the energy dissipation compared to a static one. Some authors focus on minimizing Emax, the maximum energy dissipation of any single node in the network, while others aim at minimizing Ebar, the average energy dissipation over all nodes. In our paper we take a more holistic view, considering both Emax and Ebar. The main contribution of this paper is to provide a simulation-based analysis of the energy efficiency of WSNs with static and mobile sinks. The focus is on two important configuration parameters: mobility path of the sink and duty cycling value of the nodes. On the one hand, it is well known that in the case of a mobile sink with fixed trajectory the choice of the mobility path influences energy efficiency. On the other hand, in some types of applications sensor nodes spend a rather large fraction of their total lifetime in idle mode, and therefore higher energy efficiency can be achieved by using the concept of reduced duty cycles. In particular, we quantitatively analyze the influence of duty cycling and the mobility radius of the sink as well as their interrelationship in terms of energy consumption for a well-defined model scenario. The analysis starts from general load considerations and is refined into a geometrical model. This model is validated by simulations which are more realistic in terms of duty cycling than previous work. It is illustrated that over all possible configuration scenarios in terms of duty cycle and mobility radius of the sink the energy dissipation in the WSN can vary up to a factor of nine in terms of Emax and up to a factor of 17 in terms of Ebar. It turns out that in general the choice of the duty cycle value is more important for achieving energy efficiency than the choice of the mobility radius of the sink. Moreover, for small values of the duty cycle, a static sink turns out to be optimal in terms of both Emax and Ebar. For larger values of the duty cycle, a mobile sink has advantages over a static sink, especially in terms of Emax. These insights into the basic interrelationship between duty cycle value and mobility radius of a mobile sink are relevant for energy efficient operation of homogeneous WSNs beyond our model scenario.

Congestion Avoidance and Energy Efficient Routing Protocol for Wireless Sensor Networks with aMobile Sink

Congestion severely affects the performance of a wireless sensor network in two aspects: increased data loss and reduced lifetime. This paper addresses these problems by introducing a mobile sink based routing scheme for congestion avoidance and energy efficient routing in wireless sensor networks. The proposed scheme utilizes the sink mobility and an in-network storage model that is used to set up mini-sinks along the mobility trajectory of the sink. Mini-sinks are responsible for collecting data from the sensor nodes located in their vicinity, thus avoiding data flow to a single data collection point, e.g., a static sink that is the major cause of congestion, data loss and reduced lifetime of the sensor network. Also, in the proposed scheme data only has to travel a limited number of hops to reach the nearest mini-sink which helps to improve the energy consumption of the sensor nodes. Through simulation we show the effectiveness of the given routing scheme in terms of congestion avoidance and increased lifetime of the wireless sensor network.

E-Mail Classification for Phishing Defense

We discuss a classification-based approach for filtering phishing messages in an e-mail stream. Upon arrival, various features of every e-mail are extracted. This forms the basis of a classification process which detects potentially harmful phishing messages. We introduce various new features for identifying phishing messages and rank established as well as newly introduced features according to their significance for this classification problem. Moreover, in contrast to classical binary classification approaches (spam vs. not spam), a more refined ternary classification approach for filtering e-mail data is investigated which automatically distinguishes three message types: ham (solicited e-mail), spam, and phishing. Experiments with representative data sets illustrate that our approach yields better classification results than existing phishing detection methods. Moreover, the direct ternary classification proposed is compared to a sequence of two binary classification processes. Direct one-step ternary classification is not only more efficient, but is also shown to achieve better accuracy than repeated binary classification.

Spam Filtering Based on Latent Semantic Indexing

In this paper, a study on the classification performance of a vector space model (VSM) and of latent semantic indexing (LSI) applied to the task of spam filtering is summarized. Based on a feature set used in the extremely widespread, de-facto standard spam filtering system SpamAssassin, a vector space model and latent semantic indexing are applied for classifying e-mail messages as spam or not spam. The test data sets used are partly from the official TREC 2005 data set and partly

Distributed QR factorization based on randomized algorithms

Most parallel algorithms for matrix computations assume a static network with reliable communication and thus use fixed communication schedules. However, in situations where computer systems may change dynamically, in particular, when they have unreliable components, algorithms with randomized communication schedule may be an interesting alternative. We investigate randomized algorithms based on gossiping for the distributed computation of the QR factorization. The analyses of numerical accuracy showed that known numerical properties of classical sequential and parallel QR decomposition algorithms are preserved. Moreover, we illustrate that the randomized approaches are well suited for distributed systems with arbitrary topology and potentially unreliable communication, where approaches with fixed communication schedules have major drawbacks. The communication overhead compared to the optimal parallel QR decomposition algorithm (CAQR) is analyzed. The randomized algorithms have a much higher potential for trading off numerical accuracy against performance because their accuracy is proportional to the amount of communication invested.

Multi-Level Reputation-Based Greylisting

We present the idea and implementation details of a highly effective and reliable e-mail filtering technique. At the core of the component-based architecture is a novel combination of an enhanced self-learning variant of greylisting with a reputation-based trust mechanism. These strategies provide separate feature extraction and classification components with the opportunity of utilizing the time between two delivery attempts of an e-mail message. The approach presented features a very high spam blocking rate and also minimizes the workload on the client side, as no responsibility for messages classified as spam is taken. The reputation-based trust mechanism decreases the delay in the transfer process of e-mail messages from reliable senders and also reduces the number of erroneously blocked legitimate messages.

Effects of Reduced Precision on Floating-Point SVM Classification Accuracy

Abstract There is growing interest in performing ever more complex classification tasks on mobile and embedded devices in real-time, which results in the need for e_cient implementations of the respective algorithms. Support vector machines (SVMs) represent a powerful class of nonlinear classifiers, and reducing the working precision represents a promising approach to achieving e_cient implementations of the SVM classification phase. However, the relationship between SVM classification accuracy and the arithmetic precision used is not yet su_ciently understood. We investigate this relationship in floating-point arithmetic and illustrate that often a large reduction in the working precision of the classification process is possible without loss in classification accuracy. Moreover, we investigate the adaptation of bounds on allowable SVM parameter perturbations in order to estimate the lowest possible working precision in floating-point arithmetic. Among the three representative data sets considered in this paper, none requires a precision higher than 15 bit, which is a considerable reduction from the 53 bit used in double precision floating-point arithmetic. Furthermore, we demonstrate analytic bounds on the working precision for SVMs with Gaussian kernel providing good predictions of possible reductions in the working precision without sacrificing classification accuracy.

An extension of the divide-and-conquer method for a class of symmetric block-tridiagonal eigenproblems

A divide-and-conquer method for computing eigenvalues and eigenvectors of a block-tridiagonal matrix with rank-one off-diagonal blocks is presented. The implications of unbalanced merging operations due to unequal block sizes are analyzed and illustrated with numerical examples. It is shown that an unfavorable order for merging blocks in the synthesis phase of the algorithm may lead to a significant increase of the arithmetic complexity. A strategy to determine a good merging order that is at least close to optimal in all cases is given. The method has been implemented and applied to test problems from a quantum chemistry application.

Computing Approximate Eigenpairs of Symmetric Block Tridiagonal Matrices

A divide-and-conquer method for computing approximate eigenvalues and eigenvectors of a block tridiagonal matrix is presented. In contrast to a method described earlier [W. N. Gansterer, R. C. Ward, and R. P. Muller, ACM Trans. Math. Software, 28 (2002), pp. 45--58], the off-diagonal blocks can have arbitrary ranks. It is shown that lower rank approximations of the off-diagonal blocks as well as relaxation of deflation criteria permit the computation of approximate eigenpairs with prescribed accuracy at significantly reduced computational cost compared to standard methods such as, for example, implemented in LAPACK.

On the detection and identification of botnets

We develop and discuss automated and self-adaptive systems for detecting and classifying botnets based on machine learning techniques and integration of human expertise. The proposed concept is purely passive and is based on analyzing information collected at three levels: (i) the payload of single packets received, (ii) observed access patterns to a darknet at the level of network traffic, and (iii) observed contents of TCP/IP traffic at the protocol level. We illustrate experiments based on real-life data collected with a darknet set up for this purpose to show the potential of the proposed concept for Levels (i) and (ii). As darknets cannot capture TCP/IP traffic data, we use a small spamtrap in our experiments at Level (iii). Strictly speaking, this approach for Level (iii) is not purely passive. However, traffic moving through a network could potentially be analyzed in a similar way to also obtain a purely passive system at this level.