Markus Völker

An approach to infrastructure-independent person localization with an IEEE 802.15.4 WSN

In this paper a new concept for an infrastructure independent approach to indoor person localization is outlined. It is proposed to deploy an ad-hoc wireless sensor network (WSN) by means of a pedestrian dead reckoning (PDR) unit. The deployed nodes are initialized with a PDR-determined position estimation. Furthermore, a subset of the nodes has a GPS connection and thus an accurate estimation of the own position. Within this infrastructure, persons that carry on-body sensor nodes can then be localized based on received signal strength (RSS) evaluation and extended Kalman filter (EKF) data fusion. An experimental setup with a 65 node ZigBee sensor network is described and the collected data are evaluated off-line. It is analyzed how RSS person localization performs under experimental real-world conditions. The proposed approach results in a localization accuracy on the order of a couple of meters for realistic parameter settings. Initial experiments on node deployment and a simulative evaluation of the influence of biased anchor node position estimations are presented. It is concluded that the proposed system can obtain an accuracy on the order of 3 meters without necessitating map knowledge or previously deployed infrastructure. This accuracy is sufficient for a range of person localization applications.

Drawing binary tanglegrams: an experimental evaluation

A tanglegram is a pair of trees whose leaf sets are in one-to-one correspondence; matching leaves are connected by inter-tree edges. In applications such as phylogenetics or hierarchical clustering, it is required that the individual trees are drawn crossing-free. A natural optimization problem, denoted tanglegram layout problem, is thus to minimize the number of crossings between inter-tree edges. The tanglegram layout problem is NP-hard even for complete binary trees, for general binary trees the problem is hard to approximate if the Unique Games Conjecture holds. In this paper we present an extensive experimental comparison of a new and several known heuristics for the general binary case. We measure the performance of the heuristics with a simple integer linear program and a new exact branch-and-bound algorithm. The new heuristic returns the first solution that the branch-and-bound algorithm computes (in quadratic time). Surprisingly, in most cases this simple heuristic is at least as good as the best of the other heuristics.

Constraint-based large neighborhood search for machine reassignment

This paper addresses a process-to-machine reassignment problem arising in cloud computing environments. The problem formulation has been posed as the ROADEF/EURO challenge 2012. Our presented approach is basically a large neighborhood search that iteratively improves a given solution. In each iteration only a subset of processes is considered for reassignment and the new assignments are evaluated by a constraint program. In this paper we present our general solution approach. Furthermore, we evaluate different process selection strategies and other optimization means to improve the performance on larger instances. In addition, we present a simple way to compute tight lower bounds of the necessary costs.

Link Scheduling in Local Interference Models

Choosing an appropriate interference model is crucial for link scheduling problems in sensor networks. While graph-based interference models allow for distributed and purely local coloring approaches which lead to many interesting results, a more realistic and widely agreed on model such as the signal-to-noise-plus-interference ratio (SINR) inherently makes scheduling radio transmission a non-local task, and thus impractical for the development of distributed and scalable scheduling protocols in sensor networks. In this work, we focus on interference models that are local in the sense that admissibility of transmissions only depends on local concurrent transmissions, and correct with respect to the geometric SINR model. In our analysis, we show lower bounds on the limitations that these restrictions impose an any such model as well as approximation results for greedy scheduling algorithms in a class of these models.

Simulation-Based analysis of topology control algorithms for wireless ad hoc networks

Topology control aims at optimizing throughput and energy consumption of wireless networks by adjusting transmission powers or by restricting the communication to a well-chosen subset of communication links. Over the years, a variety of topology control algorithms have been proposed. However, many of these algorithms have been mainly studied from a theoretical point of view. On the other hand, existing simulation-based studies often only compare few approaches based on rather simple simulations, e.g., abstracting from communication protocols. In this paper, we present a thorough study of a variety of topology control algorithms based on the methodology of algorithm engineering. To analyze achievable performance improvements for communication according to the IEEE 802.11g standard we use the ns-3 network simulator. In addition to analyzing the communication throughput, we also study the effects of topology control on the energy demand in the network. Based on our simulation results, we then identify properties of the computed topologies that are essential for the achieved improvements. The gained insights are finally used to motivate an extension of the well-known XTC algorithm, which enables significant performance improvements in the considered application scenario.

Efficient algorithms for distributed detection of holes and boundaries in wireless networks

We propose two novel algorithms for distributed and locationfree boundary recognition in wireless sensor networks. Both approaches enable a node to decide autonomously whether it is a boundary node, based solely on connectivity information of a small neighborhood. This makes our algorithms highly applicable for dynamic networks where nodes can move or become inoperative. We compare our algorithms qualitatively and quantitatively with several previous approaches. In extensive simulations, we consider various models and scenarios. Although our algorithms use less information than most other approaches, they produce significantly better results. They are very robust against variations in node degree and do not rely on simplified assumptions of the communication model. Moreover, they are much easier to implement on real sensor nodes than most existing approaches.

A constraint programming-based approach to a large-scale energy management problem with varied constraints

This paper addresses a large-scale power plant maintenance scheduling and production planning problem, which has been proposed by the ROADEF/EURO Challenge 2010. We develop two lower bounds for the problem: a greedy heuristic and a flow network for which a minimum cost flow problem has to be solved.Furthermore, we present a solution approach that combines a constraint programming formulation of the problem with several heuristics. The problem is decomposed into an outage scheduling and a production planning phase. The first phase is solved by a constraint program, which additionally ensures the feasibility of the remaining problem. In the second phase we utilize a greedy heuristic—developed from our greedy lower bound—to assign production levels and refueling amounts for a given outage schedule. All proposed strategies are shown to be competitive in an experimental evaluation.

Force-directed tracking in wireless networks using signal strength and step recognition

Force-directed approaches, also known as spring embedders, are widely used in the context of graph drawing and network embedding. In this paper, we study the application of these methods to signal-strength based tracking in wireless networks. The performance of the presented algorithms is evaluated based on pedestrian tracking experiments in a 60-node wireless sensor network (WSN). Additionally, we compare the outlined approach with implementations of Extended Kalman Filters (EKF) and examine similarities and distinctions between both approaches. The algorithms are developed in a 3-step process. First, we take a brief look at the trilateration problem, where a position is estimated based on a set of noisy signal strength measurements. From this we conclude how signal strengths can be translated into adequate spring forces. Subsequently, we establish a movement model by introducing additional forces between consecutive position estimates. Finally, we show how step information from a pedestrian can be used to improve the localization. Our experimental results indicate that force-directed methods offer an interesting and competitive approach to the tracking problem. Especially the possibility to easily include further information by introducing additional forces makes them very attractive. As modeled forces are not limited to linear functions, non-linear aspects such as distance estimates can be effortlessly modeled. We conclude that the application of force-directed approaches to the tracking problem offers a worthwhile direction for future research.

Straightening Drawings of Clustered Hierarchical Graphs

In this paper we deal with making drawings of clustered hierarchical graphs nicer. Given a planar graph G= (V,E) with an assignment of the vertices to horizontal layers, a plane drawing of G(with y-monotone edges) can be specified by stating for each layer the order of the vertices lying on and the edges intersecting that layer. Given these orders and a recursive partition of the vertices into clusters, we want to draw Gsuch that (i) edges are straight-line segments, (ii) clusters lie in disjoint convex regions, (iii) no edge intersects a cluster boundary twice. First we investigate fast algorithms that produce drawings of the above type if the clustering fulfills certain conditions. We give two fast algorithms with different preconditions. Second we give a linear programming (LP) formulation that always yields a drawing that fulfills the above three requirements--if such a drawing exists. The size of our LP formulation is linear in the size of the graph.