A Flow Trace Generator using Graph-based Traffic Classification Techniques

We propose a novel methodology to generate realistic network flow traces to enable systematic evaluation of network monitoring systems in various traffic conditions. Our technique uses a graph-based approach to model the communication structure observed in real-world traces and to extract traffic templates. By combining extracted and user-defined traffic templates, realistic network flow traces that comprise normal traffic and customized conditions are generated in a scalable manner. A proof-of-concept implementation demonstrates the utility and simplicity of our method to produce a variety of evaluation scenarios. We show that the extraction of templates from real-world traffic leads to a manageable number of templates that still enable accurate re-creation of the original communication properties on the network flow level.

3G network traffic sources measurement and analysis

In this paper, we show statistical analyses of several types of traffic sources in a 3G network, namely voice, video and data sources. For each traffic source type, measurements were collected in order to, on the one hand, gain better understanding of the statistical characteristics of the sources and, on the other hand, enable forecasting traffic behaviour in the network. The latter can be used to estimate service times and quality of service parameters. The probability density function, mean, variance, mean square deviation, skewness and kurtosis of the interarrival times are estimated by Wolfram Mathematica and Crystal Ball statistical tools. Based on evaluation of packet interarrival times, we show how the gamma distribution can be used in network simulations and in evaluation of available capacity in opportunistic systems. As a result, from our analyses, shape and scale parameters of gamma distribution are generated. Data can be applied also in dynamic network configuration in order to avoid potential network congestions or overflows. Copyright © 2013 John Wiley & Sons, Ltd.

Information-centric communication in mobile and wireless networks

Information-centric networking (ICN) is a new communication paradigm that has been proposed to cope with drawbacks of host-based communication protocols, namely scalability and security. In this thesis, we base our work on Named Data Networking (NDN), which is a popular ICN architecture, and investigate NDN in the context of wireless and mobile ad hoc networks. In a first part, we focus on NDN efficiency (and potential improvements) in wireless environments by investigating NDN in wireless one-hop communication, i.e., without any routing protocols. A basic requirement to initiate informationcentric communication is the knowledge of existing and available content names. Therefore, we develop three opportunistic content discovery algorithms and evaluate them in diverse scenarios for different node densities and content distributions. After content names are known, requesters can retrieve content opportunistically from any neighbor node that provides the content. However, in case of short contact times to content sources, content retrieval may be disrupted. Therefore, we develop a requester application that keeps meta information of disrupted content retrievals and enables resume operations when a new content source has been found. Besides message efficiency, we also evaluate power consumption of information-centric broadcast and unicast communication. Based on our findings, we develop two mechanisms to increase efficiency of information-centric wireless one-hop communication. The first approach called Dynamic Unicast (DU) avoids broadcast communication whenever possible since broadcast transmissions result in more duplicate Data transmissions, lower data rates and higher energy consumption on mobile nodes, which are not interested in overheard Data, compared to unicast communication. Hence, DU uses broadcast communication only until a content source has been found and then retrieves content directly via unicast from the same source. The second approach called RC-NDN targets efficiency of wireless broadcast communication by reducing the number of duplicate Data transmissions. In particular, RC-NDN is a Data encoding scheme for content sources that increases diversity in wireless broadcast transmissions such that multiple concurrent requesters can profit from each others’ (overheard) message transmissions. If requesters and content sources are not in one-hop distance to each other, requests need to be forwarded via multi-hop routing. Therefore, in a second part of this thesis, we investigate information-centric wireless multi-hop communication. First, we consider multi-hop broadcast communication in the context of rather static community networks. We introduce the concept of preferred forwarders, which relay Interest messages slightly faster than non-preferred forwarders to reduce redundant duplicate message transmissions. While this approach works well in static networks, the performance may degrade in mobile networks if preferred forwarders may regularly move away. Thus, to enable routing in mobile ad hoc networks, we extend DU for multi-hop communication. Compared to one-hop communication, multi-hop DU requires efficient path update mechanisms (since multi-hop paths may expire quickly) and new forwarding strategies to maintain NDN benefits (request aggregation and caching) such that only a few messages need to be transmitted over the entire end-to-end path even in case of multiple concurrent requesters. To perform quick retransmission in case of collisions or other transmission errors, we implement and evaluate retransmission timers from related work and compare them to CCNTimer, which is a new algorithm that enables shorter content retrieval times in information-centric wireless multi-hop communication. Yet, in case of intermittent connectivity between requesters and content sources, multi-hop routing protocols may not work because they require continuous end-to-end paths. Therefore, we present agent-based content retrieval (ACR) for delay-tolerant networks. In ACR, requester nodes can delegate content retrieval to mobile agent nodes, which move closer to content sources, can retrieve content and return it to requesters. Thus, ACR exploits the mobility of agent nodes to retrieve content from remote locations. To enable delay-tolerant communication via agents, retrieved content needs to be stored persistently such that requesters can verify its authenticity via original publisher signatures. To achieve this, we develop a persistent caching concept that maintains received popular content in repositories and deletes unpopular content if free space is required. Since our persistent caching concept can complement regular short-term caching in the content store, it can also be used for network caching to store popular delay-tolerant content at edge routers (to reduce network traffic and improve network performance) while real-time traffic can still be maintained and served from the content store.

Lung Pattern Classification for Interstitial Lung Diseases Using a Deep Convolutional Neural Network

Automated tissue characterization is one of the most crucial components of a computer aided diagnosis (CAD) system for interstitial lung diseases (ILDs). Although much research has been conducted in this field, the problem remains challenging. Deep learning techniques have recently achieved impressive results in a variety of computer vision problems, raising expectations that they might be applied in other domains, such as medical image analysis. In this paper, we propose and evaluate a convolutional neural network (CNN), designed for the classification of ILD patterns. The proposed network consists of 5 convolutional layers with 2×2 kernels and LeakyReLU activations, followed by average pooling with size equal to the size of the final feature maps and three dense layers. The last dense layer has 7 outputs, equivalent to the classes considered: healthy, ground glass opacity (GGO), micronodules, consolidation, reticulation, honeycombing and a combination of GGO/reticulation. To train and evaluate the CNN, we used a dataset of 14696 image patches, derived by 120 CT scans from different scanners and hospitals. To the best of our knowledge, this is the first deep CNN designed for the specific problem. A comparative analysis proved the effectiveness of the proposed CNN against previous methods in a challenging dataset. The classification performance (~85.5%) demonstrated the potential of CNNs in analyzing lung patterns. Future work includes, extending the CNN to three-dimensional data provided by CT volume scans and integrating the proposed method into a CAD system that aims to provide differential diagnosis for ILDs as a supportive tool for radiologists.

Analysis Methodology for Flow-level Evaluation of a Hybrid Mobile-Sensor Network

Our society uses a large diversity of co-existing wired and wireless networks in order to satisfy its communication needs. A cooper- ation between these networks can benefit performance, service availabil- ity and deployment ease, and leads to the emergence of hybrid networks. This position paper focuses on a hybrid mobile-sensor network identify- ing potential advantages and challenges of its use and defining feasible applications. The main value of the paper, however, is in the proposed analysis approach to evaluate the performance at the mobile network side given the mixed mobile-sensor traffic. The approach combines packet- level analysis with modelling of flow-level behaviour and can be applied for the study of various application scenarios. In this paper we consider two applications with distinct traffic models namely multimedia traffic and best-effort traffic.

Clinical, polysomnographic and genome-wide association analyses of narcolepsy with cataplexy: a European Narcolepsy Network study

The aim of this study was to describe the clinical and PSG characteristics of narcolepsy with cataplexy and their genetic predisposition by using the retrospective patient database of the European Narcolepsy Network (EU-NN). We have analysed retrospective data of 1099 patients with narcolepsy diagnosed according to International Classification of Sleep Disorders-2. Demographic and clinical characteristics, polysomnography and multiple sleep latency test data, hypocretin-1 levels, and genome-wide genotypes were available. We found a significantly lower age at sleepiness onset (men versus women: 23.74 ± 12.43 versus 21.49 ± 11.83, P = 0.003) and longer diagnostic delay in women (men versus women: 13.82 ± 13.79 versus 15.62 ± 14.94, P = 0.044). The mean diagnostic delay was 14.63 ± 14.31 years, and longer delay was associated with higher body mass index. The best predictors of short diagnostic delay were young age at diagnosis, cataplexy as the first symptom and higher frequency of cataplexy attacks. The mean multiple sleep latency negatively correlated with Epworth Sleepiness Scale (ESS) and with the number of sleep-onset rapid eye movement periods (SOREMPs), but none of the polysomnographic variables was associated with subjective or objective measures of sleepiness. Variant rs2859998 in UBXN2B gene showed a strong association (P = 1.28E-07) with the age at onset of excessive daytime sleepiness, and rs12425451 near the transcription factor TEAD4 (P = 1.97E-07) with the age at onset of cataplexy. Altogether, our results indicate that the diagnostic delay remains extremely long, age and gender substantially affect symptoms, and that a genetic predisposition affects the age at onset of symptoms.

Application and comparison of classification algorithms for recognition of Alzheimer's disease in electrical brain activity (EEG)

The early detection of subjects with probable Alzheimer's disease (AD) is crucial for effective appliance of treatment strategies. Here we explored the ability of a multitude of linear and non-linear classification algorithms to discriminate between the electroencephalograms (EEGs) of patients with varying degree of AD and their age-matched control subjects. Absolute and relative spectral power, distribution of spectral power, and measures of spatial synchronization were calculated from recordings of resting eyes-closed continuous EEGs of 45 healthy controls, 116 patients with mild AD and 81 patients with moderate AD, recruited in two different centers (Stockholm, New York). The applied classification algorithms were: principal component linear discriminant analysis (PC LDA), partial least squares LDA (PLS LDA), principal component logistic regression (PC LR), partial least squares logistic regression (PLS LR), bagging, random forest, support vector machines (SVM) and feed-forward neural network. Based on 10-fold cross-validation runs it could be demonstrated that even tough modern computer-intensive classification algorithms such as random forests, SVM and neural networks show a slight superiority, more classical classification algorithms performed nearly equally well. Using random forests classification a considerable sensitivity of up to 85% and a specificity of 78%, respectively for the test of even only mild AD patients has been reached, whereas for the comparison of moderate AD vs. controls, using SVM and neural networks, values of 89% and 88% for sensitivity and specificity were achieved. Such a remarkable performance proves the value of these classification algorithms for clinical diagnostics.

A computer-aided diagnostic system to characterize CT focal liver lesions: design and optimization of a neural network classifier

In this paper, a computer-aided diagnostic (CAD) system for the classification of hepatic lesions from computed tomography (CT) images is presented. Regions of interest (ROIs) taken from nonenhanced CT images of normal liver, hepatic cysts, hemangiomas, and hepatocellular carcinomas have been used as input to the system. The proposed system consists of two modules: the feature extraction and the classification modules. The feature extraction module calculates the average gray level and 48 texture characteristics, which are derived from the spatial gray-level co-occurrence matrices, obtained from the ROIs. The classifier module consists of three sequentially placed feed-forward neural networks (NNs). The first NN classifies into normal or pathological liver regions. The pathological liver regions are characterized by the second NN as cyst or "other disease." The third NN classifies "other disease" into hemangioma or hepatocellular carcinoma. Three feature selection techniques have been applied to each individual NN: the sequential forward selection, the sequential floating forward selection, and a genetic algorithm for feature selection. The comparative study of the above dimensionality reduction methods shows that genetic algorithms result in lower dimension feature vectors and improved classification performance.

A reliable, traffic-adaptive and energy-efficient link layer for wireless sensor networks

The paper presents a link layer stack for wireless sensor networks, which consists of the Burst-aware Energy-efficient Adaptive Medium access control (BEAM) and the Hop-to-Hop Reliability (H2HR) protocol. BEAM can operate with short beacons to announce data transmissions or include data within the beacons. Duty cycles can be adapted by a traffic prediction mechanism indicating pending packets destined for a node and by estimating its wake-up times. H2HR takes advantage of information provided by BEAM such as neighbour information and transmission information to perform per-hop congestion control. We justify the design decisions by measurements in a real-world wireless sensor network testbed and compare the performance with other link layer protocols.

Traffic-Adaptive and Link-Quality-Aware Communication in Wireless Sensor Networks

This paper is a summary of the main contribu- tions of the PhD thesis published in [1]. The main research contributions of the thesis are driven by the research question how to design simple, yet efficient and robust run-time adaptive resource allocation schemes within the commu- nication stack of Wireless Sensor Network (WSN) nodes. The thesis addresses several problem domains with con- tributions on different layers of the WSN communication stack. The main contributions can be summarized as follows: First, a a novel run-time adaptive MAC protocol is intro- duced, which stepwise allocates the power-hungry radio interface in an on-demand manner when the encountered traffic load requires it. Second, the thesis outlines a metho- dology for robust, reliable and accurate software-based energy-estimation, which is calculated at network run- time on the sensor node itself. Third, the thesis evaluates several Forward Error Correction (FEC) strategies to adap- tively allocate the correctional power of Error Correcting Codes (ECCs) to cope with timely and spatially variable bit error rates. Fourth, in the context of TCP-based communi- cations in WSNs, the thesis evaluates distributed caching and local retransmission strategies to overcome the perfor- mance degrading effects of packet corruption and trans- mission failures when transmitting data over multiple hops. The performance of all developed protocols are eval- uated on a self-developed real-world WSN testbed and achieve superior performance over selected existing ap- proaches, especially where traffic load and channel condi- tions are suspect to rapid variations over time.

Validation and reclassification of MGAP and GAP in hospital settings using data from the Trauma Audit and Research Network.

BACKGROUND Recently, two simple clinical scores were published to predict survival in trauma patients. Both scores may successfully guide major trauma triage, but neither has been independently validated in a hospital setting. METHODS This is a cohort study with 30-day mortality as the primary outcome to validate two new trauma scores-Mechanism, Glasgow Coma Scale (GCS), Age, and Pressure (MGAP) score and GCS, Age and Pressure (GAP) score-using data from the UK Trauma Audit and Research Network. First, an assessment of discrimination, using the area under the receiver operating characteristic (ROC) curve, and calibration, comparing mortality rates with those originally published, were performed. Second, we calculated sensitivity, specificity, predictive values, and likelihood ratios for prognostic score performance. Third, we propose new cutoffs for the risk categories. RESULTS A total of 79,807 adult (≥16 years) major trauma patients (2000-2010) were included; 5,474 (6.9%) died. Mean (SD) age was 51.5 (22.4) years, median GCS score was 15 (interquartile range, 15-15), and median Injury Severity Score (ISS) was 9 (interquartile range, 9-16). More than 50% of the patients had a low-risk GAP or MGAP score (1% mortality). With regard to discrimination, areas under the ROC curve were 87.2% for GAP score (95% confidence interval, 86.7-87.7) and 86.8% for MGAP score (95% confidence interval, 86.2-87.3). With regard to calibration, 2,390 (3.3%), 1,900 (28.5%), and 1,184 (72.2%) patients died in the low, medium, and high GAP risk categories, respectively. In the low- and medium-risk groups, these were almost double the previously published rates. For MGAP, 1,861 (2.8%), 1,455 (15.2%), and 2,158 (58.6%) patients died in the low-, medium-, and high-risk categories, consonant with results originally published. Reclassifying score point cutoffs improved likelihood ratios, sensitivity and specificity, as well as areas under the ROC curve. CONCLUSION We found both scores to be valid triage tools to stratify emergency department patients, according to their risk of death. MGAP calibrated better, but GAP slightly improved discrimination. The newly proposed cutoffs better differentiate risk classification and may therefore facilitate hospital resource allocation. LEVEL OF EVIDENCE Prognostic study, level II.

First results from a Swiss level I trauma centre participating in the UK Trauma Audit and Research Network (TARN): Prospective cohort study

QUESTIONS UNDER STUDY: Patient characteristics and risk factors for death of Swiss trauma patients in the Trauma Audit and Research Network (TARN). METHODS: Descriptive analysis of trauma patients (≥16 years) admitted to a level I trauma centre in Switzerland (September 1, 2009 to August 31, 2010) and entered into TARN. Multivariable logistic regression analysis was used to identify predictors of 30-day mortality. RESULTS: Of 458 patients 71% were male. The median age was 50.5 years (inter-quartile range [IQR] 32.2-67.7), median Injury Severity Score (ISS) was 14 (IQR 9-20) and median Glasgow Coma Score (GCS) was 15 (IQR 14-15). The ISS was >15 for 47%, and 14% had an ISS >25. A total of 17 patients (3.7%) died within 30 days of trauma. All deaths were in patients with ISS >15. Most injuries were due to falls <2 m (35%) or road traffic accidents (29%). Injuries to the head (39%) were followed by injuries to the lower limbs (33%), spine (28%) and chest (27%). The time of admission peaked between 12:00 and 22:00, with a second peak between 00:00 and 02:00. A total of 64% of patients were admitted directly to our trauma centre. The median time to CT was 30 min (IQR 18-54 min). Using multivariable regression analysis, the predictors of mortality were older age, higher ISS and lower GCS. CONCLUSIONS: Characteristics of Swiss trauma patients derived from TARN were described for the first time, providing a detailed overview of the institutional trauma population. Based on these results, patient management and hospital resources (e.g. triage of patients, time to CT, staffing during night shifts) could be evaluated as a further step.

What do brain network interactions tell about cognitive processes? How to seek causality?

Optimal adjustment of brain networks allows the biased processing of information in response to the demand of environments and is therefore prerequisite for adaptive behaviour. It is widely shown that a biased state of networks is associated with a particular cognitive process. However, those associations were identified by backward categorization of trials and cannot provide a causal association with cognitive processes. This problem still remains a big obstacle to advance the state of our field in particular human cognitive neuroscience. In my talk, I will present two approaches to address the causal relationships between brain network interactions and behaviour. Firstly, we combined connectivity analysis of fMRI data and a machine leaning method to predict inter-individual differences of behaviour and responsiveness to environmental demands. The connectivity-based classification approach outperforms local activation-based classification analysis, suggesting that interactions in brain networks carry information of instantaneous cognitive processes. Secondly, we have recently established a brand new method combining transcranial alternating current stimulation (tACS), transcranial magnetic stimulation (TMS), and EEG. We use the method to measure signal transmission between brain areas while introducing extrinsic oscillatory brain activity and to study causal association between oscillatory activity and behaviour. We show that phase-matched oscillatory activity creates the phase-dependent modulation of signal transmission between brain areas, while phase-shifted oscillatory activity blunts the phase-dependent modulation. The results suggest that phase coherence between brain areas plays a cardinal role in signal transmission in the brain networks. In sum, I argue that causal approaches will provide more concreate backbones to cognitive neuroscience.

The European Narcolepsy Network (EU-NN) database.

Narcolepsy with cataplexy is a rare disease with an estimated prevalence of 0.02% in European populations. Narcolepsy shares many features of rare disorders, in particular the lack of awareness of the disease with serious consequences for healthcare supply. Similar to other rare diseases, only a few European countries have registered narcolepsy cases in databases of the International Classification of Diseases or in registries of the European health authorities. A promising approach to identify disease-specific adverse health effects and needs in healthcare delivery in the field of rare diseases is to establish a distributed expert network. A first and important step is to create a database that allows collection, storage and dissemination of data on narcolepsy in a comprehensive and systematic way. Here, the first prospective web-based European narcolepsy database hosted by the European Narcolepsy Network is introduced. The database structure, standardization of data acquisition and quality control procedures are described, and an overview provided of the first 1079 patients from 18 European specialized centres. Due to its standardization this continuously increasing data pool is most promising to provide a better insight into many unsolved aspects of narcolepsy and related disorders, including clear phenotype characterization of subtypes of narcolepsy, more precise epidemiological data and knowledge on the natural history of narcolepsy, expectations about treatment effects, identification of post-marketing medication side-effects, and will contribute to improve clinical trial designs and provide facilities to further develop phase III trials.