Network Instances Research Articles

Predicting cardiovascular risk traits from pet myocardial perfusion imaging with deep learning

Abstract Background and aim Deep Learning has revolutionised image analysis and its implementation in cardiology is rapidly advancing. DL offers the possibility to abstract highly-complex patterns from any image in order to optimise classification and prediction tasks. At the same time, positron emission tomography (PET) represents the reference technique for quantitative evaluation of myocardial perfusion and is a powerful tool for the diagnosis of myocardial ischemia and infarction. Although pathological perfusion patterns in PET are easily recognised by expert clinicians, it is unclear whether PET images harbour complex patterns inherent to traditional cardiovascular risk traits (factors) recognisable in an individual-patient basis. Hence, we aimed to deploy Deep Learning in order to predict cardiovascular risk traits from individual quantitative PET myocardial perfusion images. Methods Data form 1180 patients evaluated through quantitative N13-ammonia PET-imaging for suspected myocardial ischemia was analysed. We implemented transfer learning with fine-tuning of individual concatenated instances of an ImageNet pre-trained convolutional neural network (ResNet-50). From each PET scan, the 3 standard quantitative polar maps (rest/stress/perfusion reserve) were used as network input. A 5-fold cross validation policy was applied to training and validation for hyperparameters optimisation. Deep Learning modelling was deployed to independently predict: sex, smoking, arterial hypertension, dyslipidemia and type 2 diabetes mellitus as cardiovascular risk traits. Final model performance was evaluated through AUC and accuracy (with associated standard deviations [SD]) on a hold-out test set of 104 scans. Deep Learning-based heat maps were generated to identify regions of interest in PET imaging related to each of the predicted risk trait Results Deep Learning was able to strongly predict sex (0.87±0.04, 78%±4) and diabetes mellitus (0.75±0.13,71%±8), while its performance was only discrete for smoking (0.63±0.11, 85%±2), hypertension (0.61±0.06,58%±3) and dyslipidemia (0.59±0.05,57%±3), albeit all statistically significant (p<0.01). Conclusion Deep Learning is able to significantly predict cardiovascular risk traits from individual quantitative PET myocardial perfusion images. This suggests the existence of complex high-dimensional and localised features within cardiac imaging that relate to cardiovascular risk factors at the individual level, which definitely warrants further research. Funding Acknowledgement Type of funding source: None

Modeling and Analysis of Information Security Threats in Wireless Sensor Networks

The paper presents results of analysis of information security threats in wireless sensor networks and means for their modeling. Research subject. The subject of the research is means for modeling and analyzing information security threats to wireless sensor networks. Method. The work uses methods of mathematical modeling and system analysis. Core results. The analysis of actual threats to information security of wireless sensor networks is performed. A comparative analysis of means for simulation modeling of wireless sensor networks and attacking influences is performed. Two models of presentation of wireless sensor networks are proposed for solving the problems of threat modeling. Practical significance. The results of the work can be used for modeling and evaluation of security of complex wireless sensor networks with a large number of nodes in conditions of organizational complexity of using full-featured instances of wireless sensor networks to solve information security problems.

A Comparative Evaluation of Nature Inspired Algorithms for Telecommunication Network Design

The subject of the study was an application of nature-inspired metaheuristic algorithms to node configuration optimization in optical networks. The main objective of the optimization was to minimize capital expenditure, which includes the costs of optical node resources, such as transponders and amplifiers used in a new generation of optical networks. For this purpose a model that takes into account the physical phenomena in the optical network is proposed. Selected nature-inspired metaheuristic algorithms were implemented and compared with a reference, deterministic algorithm, based on linear integer programming. For the cases studied the obtained results show that there is a large advantage in using metaheuristic algorithms. In particular, the evolutionary algorithm, the bees algorithm and the harmony search algorithm showed superior performance for the considered data-sets corresponding to large optical networks; the integer programming-based algorithm failed to find an acceptable sub-optimal solution within the assumed maximum computational time. All optimization methods were compared for selected instances of realistic teletransmission networks of different dimensions subject to traffic demand sets extracted from real traffic data.

Elastic Traffic Engineering Subject to a Fair Bandwidth Allocation via Bilevel Programming

The ability of TCP’s congestion control scheme to adapt the rate of traffic flows and fairly use all the available resources is one of the Internet’s pillars. So far, however, the elasticity of traffic has been disregarded in traffic engineering (TE) methodologies mainly because, only recently, the increase in access capacity has moved the bottlenecks from the access network to the operator network and hungry cloud-based applications have begun to use all the available bandwidth. We propose a new approach to TE with elastic demands which models the interaction between the network operator and the end-to-end congestion control scheme as a Stackelberg game . Given a set of elastic traffic demands only specified by their origin-destination pairs, the network operator chooses a set of routing paths (leader’s problem) which, when coupled with the fair bandwidth allocation that the congestion control scheme would determine for the chosen routing (follower’s problem), maximizes a network utility function. We present bilevel programming formulations for the above TE problem with two widely-adopted bandwidth allocation models, namely, max-min fairness and proportional fairness, and derive corresponding exact and approximate single-level mathematical programming reformulations. After discussing some key properties, we report on computational results obtained for different network topologies and instance sizes. Interestingly, even feasible solutions to our bilevel TE problems with large optimality gaps yield substantially higher network utility values than those obtained by solving a standard single-level TE problem and then fairly reallocating the bandwidth a posteriori .

Designing a novel hybrid healthcare teleconsultation network: a benchtop study of telepathology in Iran and a systematic review

BackgroundGrowing demand for medical services has increased patient waiting time due to the limited number or unbalanced distribution of healthcare centers. Healthcare teleconsultation networks are one of the potentially powerful systems to overcome this problem. Medical pathology can hugely benefit from teleconsultation networks because having second opinions is precious for many cases; however, resource planning (i.e., assignment and distribution of pathology consultation requests) is challenging due to bulky medical images of patients. This results in high setup and operational costs. The aim of this study is to design an optimal teleconsultation network for pathology labs under the supervision of medical sciences universities in Tehran, Iran.MethodsTo avoid the setup cost, we first propose a modified hybrid peer-to-peer (P2P) overlay architecture for our telepathology network, using Iran’s National Healthcare Information Network (SHAMS) as the underlying infrastructure. Then we apply optimization techniques to solve the request assignment and distribution problems in the network. Finally, we present a novel mathematical model with the objective of minimizing the variable operational costs of the system.ResultsThe efficiency of the proposed method was evaluated by a set of practical-sized network instances simulated based on the characteristics of SHAMS. The results show that the presented model and architecture can obtain optimal solutions for network instances up to 350 nodes, which covers our target network.ConclusionsWe believe that the proposed method can be beneficial for designing large-scale medical teleconsultation networks by adjusting the constraints according to the rules and conditions of each country. Our findings showed that teleconsultation networks in countries with strong information technology (IT) infrastructures are under the influence of consultation fees, while in countries with weak IT infrastructure, the transmission costs are more critical. To the best of our knowledge, no research has so far addressed resource planning in medical teleconsultation networks using optimization techniques. Besides, the target network, i.e., pathology labs under the supervision of medical sciences universities in Tehran and the SHAMS network, are discussed for the first time in this work.

Optimization of latency-aware flow allocation in NGFI networks

The next generation fronthaul interface (NGFI) architecture has been specified recently in the IEEE P1914.1 standard for packet-based fronthaul transport networks. NGFI is a flexible and cost-effective solution developed for meeting stringent latency and throughput requirements of future centralized and virtualized radio access networks (C-RAN/vRAN) in the 5th generation mobile networks (5G). NGFI allows for splitting baseband processing functions of a radio frequency signal between radio unit (RU), distributed unit (DU), and central unit (CU), where each of these elements may be located at a different site of the network. In the NGFI network, the radio data to be processed is transmitted between the RU, DU, and CU in the form of packets and routed by means of packet switches over a common packet-based fronthaul transport network. In this paper, we address a basic optimization problem in an NGFI network that concerns allocation of transmission resources in network links for the RU–DU and DU–CU data flows assuming quality of service (QoS) requirements related to maximal latency of these flows. To formulate the latency-aware flow allocation (LFA) optimization problem, we apply the mixed integer programming (MIP) approach. To generate solutions to larger LFA problem instances, we develop a meta-heuristic algorithm. To account for latency constraints in both MIP and meta-heuristic, we make use of a worst-case latency estimation model. The results of numerical experiments run in three network topologies for different network instances show the effectiveness of the meta-heuristic method, when compared to the MIP approach. These results also allow us to assess the performance of the NGFI network considered.

Large-scale influence maximization via maximal covering location

Influence maximization aims at identifying a limited set of key individuals in a (social) network which spreads information based on some propagation model and maximizes the number of individuals reached. We show that influence maximization based on the probabilistic independent cascade model can be modeled as a stochastic maximal covering location problem. A reformulation based on Benders decomposition is proposed and a relation between obtained Benders optimality cuts and submodular cuts for correspondingly defined subsets is established. We introduce preprocessing tests, which allow us to remove variables from the model and develop efficient algorithms for the separation of Benders cuts. Both aspects are shown to be crucial ingredients of the developed branch-and-cut algorithm since real-life social network instances may be very large. In a computational study, the considered variants of this branch-and-cut algorithm outperform the state-of-the-art approach for influence maximization by orders of magnitude.

Editorial: Special Issue on Compact Deep Neural Networks With Industrial Applications

The twenty-five papers in this special issue focus on industrial applications that deplpoy compact deep neural networks. Artificial neural netorks have been adopted for a broad range of tasks in areas like multimedia analysis and processing, media coding, data analytics, etc. Their recent success is based on the feasibility of processing much larger and complex neural networks (deep neural networks, DNNs) than in the past, and the availability of large-scale training data sets. As a consequence, the large memory footprint of trained neural networks and the high computational complexity of performing inference cannot be neglected. Many applications require the deployment of a particularly trained network instance, potentially to a larger number of devices, which may have limitations in terms of processing power and memory. For such applications, compact representations of neural networks are of increasing relevance. In spite of active research in the area, there are still open questions to be clarified concerning, for example, how to train neural networks with optimal performance while achieving compact representations, and how to achieve representations that do not only allow for compact transmission, but also for efficient inference. This issue sheds light on these important questions and provides answers that may be useful for both practitioners and researchers in the field.

An efficient SDN‐based LTE‐WiFi spectrum aggregation system for heterogeneous 5G networks

SummaryNowadays, the continuously increasing demand for high data traffic and providing different quality of services (QoS) to the customer are very challenging tasks for all network operators. In the last few years, mobile data traffic is increased to a significant extent and half of the traffic is provided through WiFi technology which is known as WiFi offloading. To overcome the increasing traffic demand, WiFi offloading is the best option to reduce the burden of cellular networks. So, by aggregating existing indoor WiFi technology to the cellular network increases the network capacity and provides better QoS to customers. In this article, we propose and implement the LTE‐WiFi aggregation system where eNodeB is responsible for the aggregation of the WiFi access point without modifying the core network. Furthermore, the proposed system is integrated with the mobile‐CORD (M‐CORD) platform which leverages software defined networking (SDN), network function virtualization (NFV), and cloud technologies for providing a 5G environment. M‐CORD platform has three main modules: service orchestrator (XOS), SDN controller ONOS, and OpenStack. One of the important features of M‐CORD is to provide virtualized core network functions that enable the users to automatically customize, monitor, and control the resources of the network. Due to ONOS controller support, we can easily scale up the network instances by giving the configurations to service orchestrator XOS of the M‐CORD. The implementation of the proposed system is based on the OpenAirInterface (OAI) platform which provides open sources implementation of core and access networks. The aggregation of both LTE and WiFi technologies is done at the PDCP layer in a very tight coupling way. Moreover, we test our proposed system with three kinds of policies for UDP and TCP traffic: LTE only, WiFi only, and LTE‐WiFi aggregated. The experimental results show that our proposed LTE‐WiFi aggregated system gives better performance and provides high bandwidth as compared to the LTE network.

Swarm Intelligent Data Aggregation in Wireless Sensor Network

Data aggregation in WSNs is an interesting problem wherein data sensed by the sensors is routed to an aggregation node in an efficient way. Since the sensors are battery operated, it is very important for a routing protocol to conserve energy and also ensure load balancing and faster delivery. In this study, a multi-objective linear programming model is developed for this problem and solved using an exact algorithm applying dominance principle. In order to ensure faster convergence, routing algorithms incorporating strategies of swarms in nature such as Ants, Bees and Fireflies are adapted. In the simulation study, it is quite evident from the convergence characteristics, swarm intelligent algorithms could converge earlier than the exact algorithm with convergence time lesser by 90%. Moreover, when exact algorithm could solve smaller networks, the swarm intelligent algorithms could solve even larger network instances. Firefly algorithm is able to yield approximated pareto – optimal routes which outperforms ant colony optimization and bee colony optimization algorithms.

TOPSIS‐based approach for network slice selection in 5G mobile systems

SummaryFifth‐generation (5G) mobile networks can be perceived as highly manageable systems that provide increased performance while supporting a variety of services with widely diverse requirements. Recently, network slicing has been proposed by academia and industry as a resource provisioning technique capable to meet these requirements with reduced operating costs while opening new horizons for network efficiency. The aim of network slice selection function (NSSF) is an optimal selection of network instances serving the users, based on local configuration, and other available information including radio access networks (RANs) performances in the registration area. In this paper, NSSF based on Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is proposed. Here, the network slices are treated as alternatives while their performance indicators are considered as the criteria for decision making. The influences of various alternative techniques, taking into account all the three stages of decision making process, that is, normalization, weighting, and ranking, are analyzed through the rank reversal phenomenon and computational complexity. Simulation results reveal that the proposed techniques can significantly improve the network slice selection procedure.

Optimal State Replication in Stateful Data Planes

In SDN stateful data planes, switches can execute algorithms to process traffic based on local states. This approach permits to offload decisions from the controller to the switches, thus to reduce the latency to react to network events. We consider distributed network applications that process traffic at each switch based on local replicas of network-wide states. Replicating a state across multiple switches poses many challenges, because the number of state replicas and their placement affects both the data traffic and the synchronization traffic among the replicas. In our work, we formulate the optimal placement problem for replicated states, taking into account the data traffic routing, to ensure that the network applications affect the proper traffic flows, and the synchronization traffic between replicas, to ensure state coherence. Due to the high complexity required to find the optimal solution, we propose an approximated algorithm able to scale to large network instances. We show numerically that this algorithm, despite its simplicity, approximates quite well the optimal solution. We also show the beneficial effects of replicating the states with respect to the single-replica scenario, so far considered in the literature. Finally, we provide an asymptotic analysis to find the optimal number of replicas.

Performance analysis of binary and multiclass models using azure machine learning

Network data is expanding and that too at an alarming rate. Besides, the sophisticated attack tools used by hackers lead to capricious cyber threat landscape. Traditional models proposed in the field of network intrusion detection using machine learning algorithms emphasize more on improving attack detection rate and reducing false alarms but time efficiency is often overlooked. Therefore, in order to address this limitation, a modern solution has been presented using Machine Learning-as-a-Service platform. The proposed work analyses the performance of eight two-class and three multiclass algorithms using UNSW NB-15, a modern intrusion detection dataset. 82,332 testing samples were considered to evaluate the performance of algorithms. The proposed two class decision forest model exhibited 99.2% accuracy and took 6 seconds to learn 1,75,341 network instances. Multiclass classification task was also undertaken wherein attack types like generic, exploits, shellcode and worms were classified with a recall percentage of 99%, 94.49%, 91.79% and 90.9% respectively by the multiclass decision forest model that also leapfrogged others in terms of training and execution time.

Overlapping community detection based on the union of all maximum spanning trees

PurposeThe state-of-the-art methods designed for overlapping community detection are limited by their high execution time as in CPM or the need to provide some parameters like the number of communities in Bigclam and Nise_sph, which is a nontrivial information. Hence, there is a need to develop the accuracy that represents the primordial goal, where the actual state-of-the-art methods do not succeed to achieve high correspondence with the ground truth for many instances of networks. The paper aims to discuss this issue.Design/methodology/approachThe authors offer a new method that explore the union of all maximum spanning trees (UMST) and models the strength of links between nodes. Also, each node in the UMST is linked with its most similar neighbor. From this model, the authors extract local community for each node, and then they combine the produced communities according to their number of shared nodes.FindingsThe experiments on eight real-world data sets and four sets of artificial networks show that the proposed method achieves obvious improvements over four state-of-the-art (BigClam, OSLOM, Demon, SE, DMST and ST) methods in terms of the F-score and ONMI for the networks with ground truth (Amazon, Youtube, LiveJournal and Orkut). Also, for the other networks, it provides communities with a good overlapping modularity.Originality/valueIn this paper, the authors investigate the UMST for the overlapping community detection.

Accurate link prediction method based on path length between a pair of unlinked nodes and their degree

The link prediction problem has received much attention since the beginnings of social and behavioral sciences. For instance, social networks such as Facebook, Twitter, and LinkendIN change enduringly as new connections appear in the graph. For these networks, one of the biggest challenges is to find accurately the best recommendations to the users. Within the meaning of the graph, the main objective of the link prediction problem is to predict the upcoming links from the actual state of a graph. Link prediction methods use some score functions, such as Jaccard coefficient, Katz index, and Adamic Adar metric, to measure the probability of adding the links to the network. These metrics are widely used in various applications due to their simplicity and their interpretability; however, the majority of them are designed for a specific domain. Social networks become very large with a several number of users that are connected with different kinds of links. Predicting those links is still a challenging task, as we need to find the best way to perform predictions as accurate as possible. Along this way, we extend our previous work is (Jibouni et al. in 2018 6th international conference on wireless networks and mobile communications (WINCOM). IEEE, pp 1–6, 2018) where we have proposed a new node similarity measure based on the path depth between the source and destination nodes and their degrees. The used topological features are very easy to compute and very effective in solving the link prediction problem. In addition, we verify the impact of the path length l on the method performance and we show that the proposed method provides more accurate recommendations by using the path length 2 and 3. Then, we compare 13 state-of-the-art methods against the proposed method in terms of their prediction performance using the area under curve. The results on five instances of social networks show the efficiency of the proposed method in providing accurate recommendations. Furthermore, we consider machine learning techniques such as K-nearest neighbors, logistic regression, artificial neural network, decision tree, random forest, support vector machine to solve the link prediction problem as a binary classification task. The results confirm the significant accuracy improvement that can be achieved using the proposed metric.

Community detection in networks using bio-inspired optimization: Latest developments, new results and perspectives with a selection of recent meta-heuristics

Detecting groups within a set of interconnected nodes is a widely addressed problem that can model a diversity of applications. Unfortunately, detecting the optimal partition of a network is a computationally demanding task, usually conducted by means of optimization methods. Among them, randomized search heuristics have been proven to be efficient approaches. This manuscript is devoted to providing an overview of community detection problems from the perspective of bio-inspired computation. To this end, we first review the recent history of this research area, placing emphasis on milestone studies contributed in the last five years. Next, we present an extensive experimental study to assess the performance of a selection of modern heuristics over weighted directed network instances. Specifically, we combine seven global search heuristics based on two different similarity metrics and eight heterogeneous search operators designed ad-hoc. We compare our methods with six different community detection techniques over a benchmark of 17 Lancichinetti–Fortunato–Radicchi network instances. Ranking statistics of the tested algorithms reveal that the proposed methods perform competitively, but the high variability of the rankings leads to the main conclusion: no clear winner can be declared. This finding aligns with community detection tools available in the literature that hinge on a sequential application of different algorithms in search for the best performing counterpart. We end our research by sharing our envisioned status of this area, for which we identify challenges and opportunities which should stimulate research efforts in years to come.

On greedy and strategic evaders in sequential interdiction settings with incomplete information

We consider a class of sequential network interdiction problem settings where the interdictor has incomplete initial information about the network while the evader has complete knowledge of the network including its structure and arc costs. In each decision epoch, the interdictor can block (for the duration of the epoch) at most k arcs known to him/her. By observing the evader’s actions, the interdictor learns about the network structure and costs and thus, can adjust his/her actions in subsequent decision epochs. It is known from the literature that if the evader is greedy (i.e., the shortest available path is used in each decision epochs), then under some assumptions the greedy interdiction policies that block k-most vital arcs in each epoch are efficient and have a finite regret. In this paper, we consider the evader’s perspective and explore deterministic “strategic” evasion policies under the assumption that the interdictor is greedy. We first study the theoretical computational complexity of the evader’s problem. Then we derive basic constructive properties of optimal evasion policies for two decision epochs when the interdictor has no initial information about the network structure. These properties are then exploited for the design of a heuristic algorithm for a strategic evader in a general setting with an arbitrary time horizon and any initial information available to the interdictor. Our computational experiments demonstrate that the proposed heuristic outperforms the greedy evasion policy on several classes of synthetic network instances under either perfect or noisy information feedback. Finally, some interesting insights from our theoretical and computational results conclude the paper.

Relative Assortativity Index: A Quantitative Metric to Assess the Impact of Link Prediction Techniques on Assortativity of Complex Networks

Abstract We propose a quantitative metric (called relative assortativity index, RAI) to assess the extent with which a real-world network would become relatively more assortative due to link addition(s) using a link prediction technique. Our methodology is as follows: for a link prediction technique applied on a particular real-world network, we keep track of the assortativity index values incurred during the sequence of link additions until there is negligible change in the assortativity index values for successive link additions. We count the number of network instances for which the assortativity index after a link addition is greater or lower than the assortativity index prior to the link addition and refer to these counts as relative assortativity count and relative dissortativity count, respectively. RAI is computed as (relative assortativity count − relative dissortativity count) / (relative assortativity count + relative dissortativity count). We analyzed a suite of 80 real-world networks across different domains using 3 representative neighborhood-based link prediction techniques (Preferential attachment, Adamic Adar and Jaccard coefficients [JACs]). We observe the RAI values for the JAC technique to be positive and larger for several real-world networks, while most of the biological networks exhibited positive RAI values for all the three techniques.

Integer linear programming formulations for the variable data rate and variable channel bandwidth scheduling problem in wireless networks

The IEEE 802.11ac standard enables a higher transmission speed than the previous IEEE standards because it allows the network frequency spectrum to be divided into communication channels of different bandwidths, varying from 20MHz to 160MHz. In this paper, we introduce the Variable Rate and Variable Bandwidth Scheduling Problem (VRBSP), which is a generalization of the classical Variable Rate Scheduling Problem (VRSP) on wireless networks. The best algorithm in the literature of VRSP, Datarate PPTAS, can only be efficiently run on networks with up to 64 links and does not guarantee optimality. In this paper, we propose two Mixed Integer Linear Programming (MILP) formulations that are used within a MILP solver to seek optimal schedules for VRBSP. This approach can also be used to solve VRSP. The computational experiments were carried out on classical network instances from the literature with up to 2048 links. They show that the MILP-based exact algorithms were able to find optimal VRBSP schedules for networks with up to 256 links and optimal VRSP schedules for networks with up to 1024 links within 3600 seconds of running time.

A Complex-Network Approach to the Generation of Synthetic Power Transmission Networks

The application of complex-network techniques to the analysis of power grids can provide interesting complementary insights into power studies such as the generation of synthetic power grids. Synthetic power grids are realistic, albeit nonreal, case studies that are topologically consistent with respect to real power networks. As a preliminary step to the generation of synthetic power grids, we use complex-network techniques to analyze several European power networks from a topological point of view. We observe that real topologies are markedly heterogeneous. Then, we present our main contribution: a novel model that generates synthetic spatial power networks that mimics the historical evolution of power systems by taking into account economic and technical factors. The model is articulated in two steps—the first step focused on economic efficiency in which power demand is met and the second one is targeted at increasing network robustness while achieving topological attributes. The resulting synthetic networks are topologically consistent with real power grids. The parametrical nature of the proposed model allows the generation of different instances of consistent power networks, a very interesting feature for grid generation.