In many network applications, such as the Internet and infrastructure networks, nodes fail or get congested dynamically, but tracking this information about all the nodes in a network where some dynamical processes are taking place is a fundamental problem. In this work, we study the problem of inferring the complete set of failed nodes, when only a sample of the node failures are known---we will be referring to this particular problem as prob{} . We consider the setting in which there exists correlations between node failures in networks, which has been studied in the case of many infrastructure networks. We formalize the prob{} problem using the Minimum Description Length (MDL) principle and we show that, in general, finding solutions that minimize the MDL cost is hard, and develop efficient algorithms with rigorous performance guarantees for finding near-optimal MDL cost solutions. We evaluate our methods on both synthetic and real world datasets, which includes the one from WAZE. WAZE is a crowd-sourced road navigation tool, that collects and presents the traffic incident reports. We found that the proposed greedy algorithm for this problem is able to recover $80%$, on average, of the failed nodes in a network for a given partial sample of input failures, which are sampled from the true set of failures at some predefined rate. Furthermore, we have also proved that this algorithm will find a solution that has MDL cost with an additive approximation guarantee of log(n) from the optimal. / Master of Science

The Internet of Things (IoT) is a vision for an internetwork of intelligent, communicating objects, which is on the cusp of transforming human lives. Smart transportation is one of the critical application domains of IoT and has benefitted from using state-of-the-art technology to combat urban issues such as traffic congestion while promoting communication between the vehicles, increasing driver safety, traffic efficiency and ultimately paving the way for autonomous vehicles. Connected Vehicle (CV) technology, enabled by Dedicated Short Range Communication (DSRC), has attracted significant attention from industry, academia, and government, due to its potential for improving driver comfort and safety. These vehicular communications have stringent transmission requirements. To assure the effectiveness and reliability of DRSC, efficient algorithms are needed to ensure adequate quality of service in the event of network congestion. Previously proposed congestion control methods that require high levels of cooperation among Vehicular Ad-Hoc Network (VANET) nodes. This paper proposes a new approach, VANETomo, which uses statistical Network Tomography (NT) to infer transmission delays on links between vehicles with no cooperation from connected nodes. Our proposed method combines open and closed loops congestion control in a VANET environment. Simulation results show VANETomo outperforming other congestion control strategies.

Doctor of Philosophy / Department of Electrical and Computer Engineering / Balasubramaniam Natarajan / Widely recognized security vulnerabilities in current wireless radio access technologies undermine the benefits of ubiquitous mobile connectivity. Security strategies typically rely on bit-level cryptographic techniques and associated protocols at various levels of the data processing stack. These solutions have drawbacks that have slowed down the progress of new wireless services. Physical layer security approaches derived from an information theoretic framework have been recently proposed with secret key generation being the primary focus of this dissertation. Previous studies of physical layer secret key generation (PHY-SKG) indicate that a low secret key generation rate (SKGR) is the primary limitation of this approach. To overcome this drawback, we propose novel SKG schemes to increase the SKGR as well as improve the security strength of generated secret keys by exploiting multiple input and multiple output (MIMO), cooperative MIMO (co-op MIMO) networks. Both theoretical and numerical results indicate that relay-based co-op MIMO schemes, traditionally used to enhance LTE-A network throughput and coverage, can also increase SKGR. Based on the proposed SKG schemes, we introduce innovative power allocation strategies to further enhance SKGR. Results indicate that the proposed power allocation scheme can offer 15% to 30% increase in SKGR relative to MIMO/co-op MIMO networks with equal power allocation at low-power region, thereby improving network security. Although co-op MIMO architecture can offer significant improvements in both performance and security, the concept of joint transmission and reception with relay nodes introduce new vulnerabilities. For example, even if the transmitted information is secured, it is difficult but essential to monitor the behavior of relay nodes. Selfish or malicious intentions of relay nodes may manifest as non-cooperation. Therefore, we propose relay node reliability evaluation schemes to measure and monitor the misbehavior of relay nodes. Using a power-sensing based reliability evaluation scheme, we attempt to detect selfish nodes thereby measuring the level of non-cooperation. An overall node reliability evaluation, which can be used as a guide for mobile users interested in collaboration with relay nodes, is performed at the basestation. For malicious behavior, we propose a network tomography technique to arrive at node reliability metrics. We estimate the delay distribution of each internal link within a co-op MIMO framework and use this estimate as an indicator of reliability. The effectiveness of the proposed node reliability evaluations are demonstrated via both theoretical analysis and simulations results. The proposed PHY-SKG strategies used in conjunction with node reliability evaluation schemes represent a novel cross-layer approach to enhance security of cooperative networks.

The massive growth and proliferation of media, content, and services on the Internet are driving the need for more network capacity as well as larger networks. With increasing bandwidth and transmission speeds, even small disruptions in service can result in a significant loss of data. Thus, it is becoming increasingly important to monitor networks for their performance and to be able to handle failures effectively. Doing so is beneficial from a network design perspective as well as in being able to provide a richer experience to the users of such networks. Network tomography refers to inference problems in large-scale networks wherein it is of interest to infer individual characteristics, such as link delays, through aggregate measurements, such as end-to-end path delays. In this dissertation, we establish a fundamental theory for a class of network tomography problems in which the link metrics of a network are modeled to be additive. We establish the necessary and sufficient conditions on the network topology, provide polynomial time graph algorithms that quantify the extent of identifiability, and algorithms to identify the unknown link metrics. We develop algorithms for all graph topologies classified on the basis of their connectivity. The solutions developed in this dissertation extend beyond networking and are applicable in areas such as nano-electronics and power systems. We then develop graph algorithms to handle link failures effectively and to provide multipath routing capabilities in IP as well as Ethernet based networks. Our schemes guarantee recovery and are designed to pre-compute alternate next hops that can be taken upon link failures. This allows for fast re-routing as we avoid the need to wait for (control plane) re-computations.

The measurability of Mobile ad hoc network (MANET) is the precondition of itsmanagement, performance optimization and network resources re-allocations. However, MANET is an infrastructure-free, multi-hop, andself-organized temporary network, comprised of a group of mobile nodes with wirelesscommunication devices. Not only does its topology structure vary with time going by, butalso the communication protocol used in its network layer or data link layer is diverse andnon-standard.In order to solve the problem of interior links performance (such as packet loss rate anddelay) measurement in MANET, this thesis has adopted an external measurement basedon network tomography (NT). To the best of our knowledge, NT technique is adaptable for Ad Hoc networkmeasurement.This thesis has deeply studied MANET measurement technique based on NT. The maincontributions are:(1) An analysis technique on MANET topology dynamic characteristic based onmobility model was proposed. At first, an Ad Hoc network mobility model formalizationis described. Then a MANET topology snapshots capturing method was proposed to findand verify that MANET topology varies in steady and non-steady state in turnperiodically. At the same time, it was proved that it was practicable in theory to introduceNT technique into Ad Hoc network measurement. The fitness hypothesis verification wasadopted to obtain the rule of Ad Hoc network topology dynamic characteristic parameters,and the Markov stochastic process was adopted to analyze MANET topology dynamiccharacteristic. The simulation results show that the method above not only is valid andgenerable to be used for all mobility models in NS-2 Tool, but also could obtain thetopology state keeping experimental formula and topology state varying probabilityformula.IV(2) An analysis technique for MANET topology dynamic characteristic based onmeasurement sample was proposed. When the scenario file of mobile models could notbe obtained beforehand, End-to-End measurement was used in MANET to obtain thepath delay time. Then topology steady period of MANET is inferred by judging whetherpath delay dithering is close to zero. At the same time, the MANET topology wasidentified by using hierarchical clustering method based on measurement sample of pathperformance during topology steady period in order to support the link performanceinference. The simulation result verified that the method above could not only detect themeasurement window time of MANET effectively, but also identify the MANETtopology architecture during measurement window time correctly.(3) A MANET link performance inference algorithm based on linear analysis modelwas proposed. The relation of inequality between link and path performance, such as lossrate of MANET, was deduced according to a linear model. The phenomena thatcommunication characteristic of packets, such as delay and loss rate, is more similarwhen the sub-paths has longer shared links was proved in the document. When the rankof the routing matrix is equal to that of its augmentation matrix, the linear model wasused to describe the Ad Hoc network link performance inference method. The simulationresults show that the algorithm not only is effective, but also has short computing time.(4) A Link performance inference algorithm based on multi-objectives optimizationwas proposed. When the rank of the routing matrix is not equal to that of its augmentationmatrix, the link performance inference was changed into multi-objectives optimizationand genetic algorithm is used to infer link performance. The probability distribution oflink performance in certain time t was obtained by performing more measurements andstatistically analyzing the hypo-solutions. Through the simulation, it can be safelyconcluded that the internal link performance, such as, link loss ratio and link delay, can beinferred correctly when the rank of the routing matrix is not equal to that of itsaugmentation matrix.

abstract The parameters required for network monitoring are not directly measurable and could be estimated indirectly by network tomography. Some important research issues, related to network tomography, motivated the research in this dissertation. The research work in this dissertation makes four significant novel contributions to the field of network tomography. These research contributions were focused on the blind techniques for performing network tomography, the modeling of errors in network tomography, improving estimates with multi-metric-based network tomography, and distributed network tomography. All of these four research problems, related to network tomography, were solved by various blind techniques including NNMF, SCS, and NTF. These contributions have been verified by processing the data obtained from laboratory experiments and by examining the correlation between the estimated and measured link delays. Evaluation of these contributions was based on the data obtained from various test beds that consisted of networking devices.

Computer networks are becoming increasingly large and complex; more so with the recent penetration of the internet into all walks of life. It is essential to be able to monitor and to analyse networks in a timely and efficient manner; to extract important metrics and measurements and to do so in a way which does not unduly disturb or affect the performance of the network under test. Network tomography is one possible method to accomplish these aims. Drawing upon the principles of statistical inference, it is often possible to determine the statistical properties of either the links or the paths of the network, whichever is desired, by measuring at the most convenient points thus reducing the effort required. In particular, bottleneck-link detection methods in which estimates of the delay distributions on network links are inferred from measurements made at end-points on network paths, are examined as a means to determine which links of the network are experiencing the highest delay. Initially two published methods, one based upon a single Gaussian distribution and the other based upon the method-of-moments, are examined by comparing their performance using three metrics: robustness to scaling, bottleneck detection accuracy and computational complexity. Whilst there are many published algorithms, there is little literature in which said algorithms are objectively compared. In this thesis, two network topologies are considered, each with three configurations in order to determine performance in six scenarios. Two new estimation methods are then introduced, both based on Gaussian mixture models which are believed to offer an advantage over existing methods in certain scenarios. Computationally, a mixture model algorithm is much more complex than a simple parametric algorithm but the flexibility in modelling an arbitrary distribution is vastly increased. Better model accuracy potentially leads to more accurate estimation and detection of the bottleneck. The concept of increasing flexibility is again considered by using a Pearson type-1 distribution as an alternative to the single Gaussian distribution. This increases the flexibility but with a reduced complexity when compared with mixture model approaches which necessitate the use of iterative approximation methods. A hybrid approach is also considered where the method-of-moments is combined with the Pearson type-1 method in order to circumvent problems with the output stage of the former. This algorithm has a higher variance than the method-of-moments but the output stage is more convenient for manipulation. Also considered is a new approach to detection algorithms which is not dependant on any a-priori parameter selection and makes use of the Kullback-Leibler divergence. The results show that it accomplishes its aim but is not robust enough to replace the current algorithms. Delay estimation is then cast in a different role, as an integral part of an algorithm to correlate input and output streams in an anonymising network such as the onion router (TOR). TOR is used by users in an attempt to conceal network traffic from observation. Breaking the encryption protocols used is not possible without significant effort but by correlating the un-encrypted input and output streams from the TOR network, it is possible to provide a degree of certainty about the ownership of traffic streams. The delay model is essential as the network is treated as providing a pseudo-random delay to each packet; having an accurate model allows the algorithm to better correlate the streams.

Modern computer networks pose a great challenge for monitoring the network performance due to their large scale and high complexity. Directly measuring the performance of internal network elements is prohibitive due to the tremendous overhead. Alternatively, network tomography, a technique that infers the unobserved network characteristics (e.g., link delays) from a small number of measurements (e.g., end-to-end path delays), is a promising solution for monitoring the internal network state in an e cient and e ective manner. This thesis initiates two variants of network tomography: concentrated network tomography and bound-based network tomography. The former is motivated by the practical needs that network operators normally concentrate on the performance of critical paths; the latter is due to the need of estimating performance bounds whenever exact performance values cannot be determined. This thesis tackles core technical di culties in concentrated network tomography and bound- based network tomography, including (1) the path identi ability problem and the monitor deploy- ment strategy for identifying a set of target paths, (2) strategies for controlling the total error bound as well as the maximum error bound over all network links, and (3) methods of constructing measure- ment paths to obtain the tightest total error bound. We evaluate all the solutions with real-world Internet service provider (ISP) networks. The theoretical results and the algorithms developed in this thesis are directly applicable to network performance management in various types of networks, where directly measuring all links is practically impossible. / Graduate

Nowadays, the traffic tomography represents an integral component in converged networks and systems for detecting their behavioral characteristics. The dissertation deals with research of its implementation with the use of evolutionary algorithms. The research was mainly focused on innovation and solving behavioral detection data flows in networks and network anomalies using tomography and evolutionary algorithms. Within the dissertation has been proposed a new algorithm, emerging from the basics of the statistical method survival analysis, combined with a genetics’ algorithm. The proposed algorithm was tested in a model of a self-created network probe using the Python programming language and Cisco laboratory network devices. Performed tests have shown the basic functionality of the proposed solution.

Les réseaux de capteurs sans fils et mobiles constituent un champ de recherche dans lequel un grand nombre de capteurs de faible coût sont déployés dans un environnement pour observer un ou plusieurs phénomènes. Ces capteurs sont autonomes, communicant et disposent d'une réserve d'énergie limitée. Les problèmes issus de ce type de système sont nombreux : gestion de l'énergie, couverture, fusion de donnée, ...L'approche proposée dans cette thèse repose sur l'hypothèse que les réseaux de capteurs doivent exhiber des propriétés d'auto-organisation et d'autonomie. Chaque capteur est en soit autonome et peut interagir avec d'autres capteurs ce qui forme une organisation complexe. Ces capteurs ont un but à accomplir et le système possède les caractéristiques suivantes : i. le but du réseau ne peut généralement pas être résolu par un capteur uniqueii. Les capteurs doivent collaborer pour accomplir le but ou contribuer à une partie de ce but.iii. Chaque capteur réagit à son environnement en fonction de ses perceptions qui sont forcément locales et limitées.Après une introduction qui décrit le domaine et pose la problématique un état de l'art du domaine est présenté au chapitre 2. Deux contributions sont abordées dans cette thèse. D'une part, l'analyse des propriétés dynamiques de topologie des réseaux de capteurs sans fil et d'autre part la performance des liens de ce type de réseaux. Pour la topologie deux approches sont proposées : au chapitre 3 une première approche basée sur le modèle de mobilité et au chapitre 4 une approche basée sur des techniques de mesures. Pour la performance des liens, deux approches sont également proposées. La première, décrite dans le chapitre 5, est basée sur un modèle d'analyse linéaire. La deuxième, décrite au chapitre 6, repose sur une technique d'optimisation multi-objectif. / The measurability of Mobile ad hoc network (MANET) is the precondition of itsmanagement, performance optimization and network resources re-allocations. However, MANET is an infrastructure-free, multi-hop, andself-organized temporary network, comprised of a group of mobile nodes with wirelesscommunication devices. Not only does its topology structure vary with time going by, butalso the communication protocol used in its network layer or data link layer is diverse andnon-standard.In order to solve the problem of interior links performance (such as packet loss rate anddelay) measurement in MANET, this thesis has adopted an external measurement basedon network tomography (NT). To the best of our knowledge, NT technique is adaptable for Ad Hoc networkmeasurement.This thesis has deeply studied MANET measurement technique based on NT. The maincontributions are:(1) An analysis technique on MANET topology dynamic characteristic based onmobility model was proposed. At first, an Ad Hoc network mobility model formalizationis described. Then a MANET topology snapshots capturing method was proposed to findand verify that MANET topology varies in steady and non-steady state in turnperiodically. At the same time, it was proved that it was practicable in theory to introduceNT technique into Ad Hoc network measurement. The fitness hypothesis verification wasadopted to obtain the rule of Ad Hoc network topology dynamic characteristic parameters,and the Markov stochastic process was adopted to analyze MANET topology dynamiccharacteristic. The simulation results show that the method above not only is valid andgenerable to be used for all mobility models in NS-2 Tool, but also could obtain thetopology state keeping experimental formula and topology state varying probabilityformula.IV(2) An analysis technique for MANET topology dynamic characteristic based onmeasurement sample was proposed. When the scenario file of mobile models could notbe obtained beforehand, End-to-End measurement was used in MANET to obtain thepath delay time. Then topology steady period of MANET is inferred by judging whetherpath delay dithering is close to zero. At the same time, the MANET topology wasidentified by using hierarchical clustering method based on measurement sample of pathperformance during topology steady period in order to support the link performanceinference. The simulation result verified that the method above could not only detect themeasurement window time of MANET effectively, but also identify the MANETtopology architecture during measurement window time correctly.(3) A MANET link performance inference algorithm based on linear analysis modelwas proposed. The relation of inequality between link and path performance, such as lossrate of MANET, was deduced according to a linear model. The phenomena thatcommunication characteristic of packets, such as delay and loss rate, is more similarwhen the sub-paths has longer shared links was proved in the document. When the rankof the routing matrix is equal to that of its augmentation matrix, the linear model wasused to describe the Ad Hoc network link performance inference method. The simulationresults show that the algorithm not only is effective, but also has short computing time.(4) A Link performance inference algorithm based on multi-objectives optimizationwas proposed. When the rank of the routing matrix is not equal to that of its augmentationmatrix, the link performance inference was changed into multi-objectives optimizationand genetic algorithm is used to infer link performance. The probability distribution oflink performance in certain time t was obtained by performing more measurements andstatistically analyzing the hypo-solutions. Through the simulation, it can be safelyconcluded that the internal link performance, such as, link loss ratio and link delay, can beinferred correctly when the rank of the routing matrix is not equal to that of itsaugmentation matrix.