Accurate estimation of the signal power received at a given location can be obtained through Ray Tracing (RT) propagation model. Nevertheless, processes involved in RT are extremely complex and not adapted for Radio Frequency (RF) mapping purposes. Particularly, the reception test process, responsible for validating received rays at given locations, contributes to over 98% of the overall RT complexity in large-scale scenarios. In this paper, we introduce a novel RT algorithm able to divide by a factor of 50 the RF map rendering time while limiting the difference in bitrate estimation to less than 1Mbps on average as compared to existing approaches. Our algorithm integrates hardware acceleration techniques for RT, proposes a new modeling of the terrain in the form of a triangular mesh, and introduces a new reception test able to efficiently calculate the signal power received per each triangle of the mesh. Our simulations show how these processes together can reduce the complexity of RF map generation, hence allowing the algorithm to scale to large-scale and precise scenarios.

Despite recent advancements in terms of network performance, end users still face slow web browsing situations, which can have a range of causes, such as a congested Wi-Fi, a bad wireless signal, or a loaded network or end host. It is thus crucial to monitor the network and troubleshoot the specific causes of slow web browsing, as this benefits end users, operators, and internet service providers alike. Various tools attempting to actively troubleshoot the network through the injection of probes exist. However, these tools are, on the one hand, expensive to run and, on the other hand, not general enough to be able to identify the specific cause of web browsing slowness. This paper addresses the problem by proposing a new lightweight passive measurement solution capable of transforming the web performance measurements collected from within the browser into indicators of network performance anomalies. We validate our solution by emulating a controlled network environment with manually injected anomalies; then, we leverage the measurement data available within the browser to build a predictive model that uses a random forest classifier to correctly classify the causes of web browsing performance degradation, with an accuracy of over 95%. This implies that one can build on our solution to propose a tool, in the form of a browser extension, that can be used in the wild to monitor the network and shed light on its anomalies by solely relying on a regular user’s web activity.

Given the dominance of adaptive video streaming services on the Internet traffic, understanding how YouTube Quality of Experience (QoE) relates to real 4G and 5G Channel Level Metrics (CLM) is of interest to not only the research community but also to Mobile Network Operators (MNOs) and content creators. In this context, we collect YouTube and CLM logs with 1-second granularity spanning a six-month period. We group the traces by their context, i.e., Mobility, Pedestrian, Bus/Railway terminals, and Static Outdoor, and derive key performance footprints of real 4G and 5G video streaming in the wild. We also develop Machine Learning (ML) classifiers to predict objective QoE video stalls by using past patterns from CLM traces. We release all datasets and software artifacts for reproducibility purposes.

Ray Tracing is an electromagnetic wave propagation modeling approach used for accurate generation of Quality of Service (QoS) maps in mobile networks. Due to its complexity, current implementation of Ray Tracing fails to generate such maps in wide areas. In this paper, we propose an optimization to Ray Tracing able to accurately generate QoS maps in a reasonable time. Using a site-specific ray launching technique and an alternative to the reception test process, we divide by almost 1200 the execution time of Ray Tracing with less than 2% of memory usage as compared to baseline solutions.

Monitoring network performance becomes crucial today since it allows content providers to ensure a good quality of their services by identifying the root causes of service degradation. Also, it gives the end-user a better understanding of the performance they have (state of the networks). A widely used monitoring technique involves performing measurements from within the browser in an effort to capture the network status as close as possible; we talk about Web-based network monitoring. Many Web measurement tools have recently been proposed, however, most of these tools either have a high computational cost or exaggeratedly consume data. In this paper, we propose a lightweight solution able to estimate the underlying network status accurately and perform Web troubleshooting in order to detect anomalies. We develop and implement a distributed system that collects measurements at both levels: browser and network. Then, we build an original network monitoring framework based on Bayesian Gaussian Mixture Models (BGMM) coupled with an algorithm to detect in real time the occurrence of anomalies. We follow a browser-based passive measurement and data-driven approach to derive our inference models, which leads to an efficient Web browsing troubleshooting solution.

Ray Tracing is a propagation modelling approach that accurately estimates the signal power received by end users while taking into account the details of the environment in their vicinity. This accuracy is at the cost of high computational load and high memory consumption due to the heavy computation performed by processes such as Ray Generation. In this paper, we introduce a site-specific ray generation technique able to generate up to 1 million rays within 5 seconds and a root mean square error for bandwidth estimation within 2 Mbps. Depending on the location of the antenna and the coverage area, our technique gives the minimum possible number of rays required in order to estimate end-users' signal power received and their download bitrate.

Web performance anomalies (e.g. time periods when metrics like page load time are abnormally high) have significant impact on user experience and revenues of web service providers. Existing methods to automatically detect web performance anomalies focus on popular websites (e.g. with tens of thousands of visits per minute). Across a wider diversity of websites, however, the number of visits per hour varies enormously, and some sites will only have few visits per hour. Low rates of visits create measurement gaps and noise that prevent the use of existing methods. This paper develops WMF, a web performance anomaly detection method applicable across a range of websites with highly variable measurement volume. To demonstrate our method, we leverage data from a website monitoring company, which allows us to leverage cross-site measurements. WMF uses matrix factorization to mine patterns that emerge from a subset of the websites to "fill in" missing data on other websites. Our validation using both a controlled website and synthetic anomalies shows that WMF's F1-score is more than double that of the state-of-the-art method. We then apply WMF to three months of web performance measurements to shed light on performance anomalies across a variety of 125 small to medium websites.

Video streaming is without doubt the most requested Internet service, and main source of pressure on the Internet infrastructure. At the same time, users are no longer satisfied by the Internet’s best effort service, instead, they expect a seamless service of high quality from the side of the network. As result, Internet Service Providers (ISP) engineer their traffic so as to improve their end-users’ experience and avoid economic losses. Content providers from their side, and to enforce customers privacy, have shifted towards end-to-end encryption (e.g., TLS/SSL). Video streaming relies on the dynamic adaptive streaming over HTTP protocol (DASH) which takes into consideration the underlying network conditions (e.g., delay, loss rate, and throughput) and the viewport capacity (e.g., screen resolution) to improve the experience of the end user in the limit of the available network resources. In this work, we propose an experimental framework able to infer fine-grained video flow information such as chunk sizes from encrypted YouTube video traces. We also present a novel technique to separate video and audio chunks from encrypted traces based on Gaussian Mixture Models (GMM). Then, we leverage our dataset to train models able to predict the class of viewport (eitherSDorHD) per video session with an average 92% accuracy and 85% F1-score. The prediction of the exact viewport resolution is also possible but shows a lower accuracy than the viewport class.

In this article, we introduce a novel dynamic controller assignment algorithm targeting connected vehicle services and applications, also known as Internet of Vehicles (IoV). The proposed approach considers a hierarchically distributed control plane, decoupled from the data plane, and uses vehicle location and control traffic load to perform controller assignment dynamically. We model the dynamic controller assignment problem as a multi-agent Markov game and solve it with cooperative multi-agent deep reinforcement learning. Simulation results using real-world vehicle mobility traces show that the proposed approach outperforms existing ones by reducing control delay as well as packet loss.

Terrestrial infrastructure-based wireless networks do not always guarantee their resources will be shared uniformly by nodes in vehicular networks. This is due mainly to the uneven and dynamic geographical distribution of vehicles and path loss effects. In this paper, we leverage multiple fifth-generation (5G) unmanned aerial vehicles (UAVs) to enhance fairness in network resource allocation among vehicles by positioning UAVs on-demand as “flying communication infrastructure”. We propose a deep reinforcement learning (DRL) approach to determine UAVs’ position to improve network resource allocation fairness and efficiency while considering the UAVs’ flying range, communication range, and energy constraints. We use a parametric fairness function to attain a number of resource allocation objectives ranging from maximizing the total throughput of vehicles, maximizing minimum throughput, and achieving proportional bandwidth allocation. Simulation results show that the proposed DRL approach to UAV positioning can improve network resource allocation according to the targeted fairness objective.

Web browsing remains one of the dominant applications of the internet, so inferring network performance becomes crucial for both users and providers (access and content) so as to be able to identify the root cause of any service degradation. Recent works have proposed several network troubleshooting tools, e.g, NDT, MobiPerf, SpeedTest, Fathom. Yet, these tools are either computationally expensive, less generic or greedy in terms of data consumption. The main purpose of this work is to leverage passive measurements freely available in the browser and machine learning techniques (ML) to infer network performance (e.g., delay, bandwidth and loss rate) without the addition of new measurement overhead. To enable this inference, we propose a framework based on extensive controlled experiments where network configurations are artificially varied and the Web is browsed, then ML is applied to build models that estimate the underlying network performance. In particular, we contrast classical ML techniques (such as random forest) to deep learning models trained using fully connected neural networks and convolutional neural networks (CNN). Results of our experiments show that neural networks have a higher accuracy compared to classical ML approaches. Furthermore, the model accuracy improves considerably using CNN.

The Internet has changed drastically in recent years, multiple novel applications and services have emerged, all about consuming digital content. In parallel, users are no longer satisfied by the Internet's best effort service, instead, they expect a seamless service of high quality from the side of the network. This has increased the pressure on Internet service providers (ISP) in their effort to efficiently engineer their traffic and improve their end-users' experience. Content providers from their side, and to further protect the content of their customers, have shifted towards end-to-end encryption (e.g., TLS/SSL), which has complicated even further the task of ISPs in handling the traffic in their network. The challenge is notable for video streaming traffic which is driving the Internet traffic growth, and which imposes tight constraints on the quality of service provided by the network depending on the content of the video stream and the equipment on the end-user premises. Video streaming relies on the dynamic adaptive streaming over HTTP (DASH) protocol which takes into consideration the underlying network conditions (e.g., delay, loss rate, and throughput) and the viewport capacity (e.g., screen resolution) to improve the experience of the end user in the limit of available resources. Nevertheless, knowing the reality of the encrypted video traffic is of great help to ISPs as it allows taking appropriate network management actions. In this work, we propose an experimental framework able to infer fine-grained video flow information such as chunk sizes from encrypted YouTube video traces. We also present a novel technique to separate video and audio chunks from encrypted traces based on Gaussian Mixture Models (GMM). We evaluate our technique with real chunk sizes (Audio/Video) collected through the browser using the Chrome Web Request API [1]. Then, we leverage these results and our dataset to train a model able to predict the class of viewport (either SD or HD) per video session with an average 92% accuracy and 85% F1 score.

The ability to monitor web and network performance becomes crucial to understand the reasons behind any service degradation. Such monitoring is also helpful to understand the relationship between the quality of experience of end users and the underlying network performance. Many troubleshooting tools have been proposed recently. They mainly consist of conducting active network measurements from within the browser. However, most of these tools either lack accuracy, or perform measurements to a limited set of servers. They are also known to introduce non-negligible overhead onto the network. The objective of this paper is to propose a new approach based on passive measurements freely available from within the web browser, and to couple these measurements to deep learning models to estimate the latency and bandwidth metrics of the underlying network without injecting any additional measurement traffic. We develop and implement our approach, and compare its estimation accuracy with the best known web-based network measurement techniques available nowadays. We follow a controlled experimental approach to derive our inference models. The results of our study show that our approach can give a very good accuracy compared to others, its accuracy is even higher than most standard techniques, and very close to the rest.

Densification through small cells and caching in base stations have been proposed to deal with the increasing demand for Internet content and the related overload on the cellular infrastructure. However, these solutions are expensive to install and maintain. Instead, using vehicles acting as mobile caches might represent an interesting alternative. In our work, we assume that users can query nearby vehicles for some time, and be redirected to the cellular infrastructure when the deadline expires. Beyond reducing costs, in such an architecture, through vehicle mobility, a user sees a much larger variety of locally accessible content within only few minutes. Unlike most of the related works on delay tolerant access, we consider the impact on the user experience by assigning different retrieval deadlines per content. In our paper, we provide the following contributions: (i) we model analytically such a scenario; (ii) we formulate an optimization problem to maximize the traffic offloaded while ensuring user experience guarantees; (iii) we propose two variable deadline policies; (iv) we perform realistic trace-based simulations, and we show that, even with low technology penetration rate, more than 60 percent of the total traffic can be offloaded which is around 20 percent larger compared to existing allocation policies.

Screen resolution along with network conditions are main objective factors impacting the user experience, in particular for video streaming applications. Terminals on their side feature more and more advanced characteristics resulting in different network requirements for good visual experience [1]. Previous studies tried to link MOS (Mean Opinion Score) to video bit rate for different screen types (e.g., CIF, QCIF, and HD) [2]. We leverage such studies and formulate a QoE-driven resource allocation problem to pinpoint the optimal bandwidth allocation that maximizes the QoE (Quality of Experience) over all users of a provider located behind the same bottleneck link, while accounting for the characteristics of the screens they use for video playout. For our optimization problem, QoE functions are built using curve fitting on data sets capturing the relationship between MOS, screen characteristics, and bandwidth requirements. We propose a simple heuristic based on Lagrangian relaxation and KKT (Karush Kuhn Tucker) conditions for a subset of constraints. Numerical simulations show that the proposed heuristic is able to increase overall QoE up to 20% compared to an allocation with TCP look-alike strategies implementing max-min fairness. Later, we use a MPEG/DASH implementation in the context of ns-3 and show that coupling our approach with a rate adaptation algorithm (e.g., [3]) can help increasing QoE while reducing both resolution switches and number of interruptions.

Video streaming is a dominant contributor to the global Internet traffic. Consequently, monitoring video streaming Quality of Experience (QoE) is of paramount importance to network providers. Monitoring QoE of video is a challenge as most of the video traffic of today is encrypted. In this paper, we consider this challenge and present an approach based on controlled experimentation and machine learning to estimate QoE from encrypted video traces using network level measurements only. We consider a case of YouTube and play out a wide range of videos under realistic network conditions to build ML models (classification and regression) that predict the subjective MOS (Mean Opinion Score) based on the ITU P.1203 model along with the QoE metrics of startup delay, quality (spatial resolution) of playout and quality variations, and this is using only the underlying network Quality of Service (QoS) features. We comprehensively evaluate our approach with different sets of input network features and output QoE metrics. Overall, our classification models predict the QoE metrics and the ITU MOS with an accuracy of 63-90% while the regression models show low error; the ITU MOS (1-5) and the startup delay (in seconds) are predicted with a root mean square error of 0.33 and 2.66 respectively.

Caching content at the edge of mobile networks is considered as a promising way to deal with the data tsunami. In addition to caching at fixed base stations or user devices, it has been recently proposed that an architecture with public or private transportation acting as mobile relays and caches might be a promising middle ground. While such mobile caches have mostly been considered in the context of delay tolerant networks, in this paper we argue that they could be used for low cost video streaming without the need to impose any delay on the user. Users can prefetch video chunks into their playout buffer from encountered vehicle caches (at low cost) or stream from the cellular infrastructure (at higher cost) when their playout buffer empties while watching the content. Our main contributions are: (i) to model the playout buffer in the user device and analyze its idle periods which correspond to bytes downloaded from the infrastructure; (ii) to optimize the content allocation to mobile caches; and to minimize the expected number of non-offloaded bytes. We perform trace-based simulations to support our findings showing that up to 60 percent of the original traffic could be offloaded from the main infrastructure.

Predicting Quality of Experience (QoE) of end users from available network Quality of Service (QoS) measurements is of significant importance for today's network and content providers. This can be achieved by using application-specific QoE models that map the network QoS to the output QoE. QoS-QoE models can be built by training supervised machine learning algorithms with training data consisting of the mappings of the input network QoS to the output QoE. In most ML works on QoE modeling, the training data is usually gathered in the wild inside the core of the service or the content provider networks. However, such data is not easily accessible to the general research community. Consequently, the training data if not available before hand, needs to be built up by controlled experimentation. Here, the fundamental challenge is the sheer amount of time consumed in collecting the datasets needed to model the QoE. Considering this problem, we present here a framework of controlled experimentation based on active learning, that allows collecting rich datasets covering the experimental space intelligently. We perform a rigorous analysis of our approach and demonstrate the performance improvement over conventional pool based uncertainty sampling for a particular use case of YouTube video streaming.

Video streaming is a dominant contributor to the global Internet traffic. Consequently, gauging network performance w.r.t. the video Quality of Experience (QoE) is of paramount importance to both telecom operators and regulators. Modern video streaming systems, e.g. YouTube, have huge catalogs of billions of different videos that vary significantly in content type. Owing to this difference, the QoE of different videos as perceived by end users can vary for the same network Quality of Service (QoS). In this paper, we present a methodology for benchmarking performance of mobile operators w.r.t Internet video that considers this variation in QoE. We take a data-driven approach to build a predictive model using supervised machine learning (ML) that takes into account a wide range of videos and network conditions. To that end, we first build and analyze a large catalog of YouTube videos. We then propose and demonstrate a framework of controlled experimentation based on active learning to build the training data for the targeted ML model. Using this model, we then devise YouScore, an estimate of the percentage of YouTube videos that may play out smoothly under a given network condition. Finally, to demonstrate the benchmarking utility of YouScore, we apply it on an open dataset of real user mobile network measurements to compare performance of mobile operators for video streaming.

Densification through small cells and caching in base stations have been proposed to deal with the increasing demand for Internet content and the related overload on the cellular infrastructure. However, these solutions are expensive to install and maintain. Instead, using vehicles acting as mobile caches might represent an interesting alternative. In our work, we assume that users can query nearby vehicles for some time, and be redirected to the cellular infrastructure when the deadline expires. Beyond reducing costs, in such an architecture, through vehicle mobility, a user sees a much larger variety of locally accessible content within only few minutes. Unlike most of the related works on delay tolerant access, we consider the impact on the user experience by assigning different retrieval deadlines per content. In our paper, we provide the following contributions: (i) we model analytically such a scenario; (ii) we formulate an optimization problem to maximize the traffic offloaded while ensuring user experience guarantees; (iii) we propose a variable deadline policy; (iv) we perform realistic trace-based simulations, and we show that, even with low technology penetration rate, more than 60% of the total traffic can be offloaded which is around 20% larger compared to existing allocation policies.

The increasing demand for mobile data is overloading the cellular infrastructure. Small cells and edge caching is being explored as an alternative, but installation and maintenance costs for sufficient coverage are significant. In this work, we perform a preliminary study of an alternative architecture based on two main ideas: (i) using vehicles as mobile caches that can be accessed by user devices; compared to small cells, vehicles are more widespread and require lower costs; (ii) combining the mobility of vehicles with delayed content access to increase the number of cache hits (and reduce the load on the infrastructure). Contrary to standard DTN-type approaches, in our system max delays are guaranteed to be kept to a few minutes (beyond this deadline, the content is fetched from the infrastructure). We first propose an analytical framework to compute the optimal number of content replicas that one should cache, in order to minimize the infrastructure load. We then investigate how to optimally refresh these caches to introduce new contents, as well as to react to the temporal variability in content popularity. Simulations suggest that our vehicular cloud considerably reduces the infrastructure load in urban settings, assuming modest penetration rates and tolerable content access delays.

In the last decade, path discovery has been extensively covered in the literature. In its simplest form, it generally works by sending probes that expire along the path from a host to a destination. It is also known that network administrators often configure their routers to limit the amount of ICMP replies sent, a common practice typically referred to as ICMP rate limitation. In this paper we attempt to characterize the responsiveness of routers to expiring ICMP echo-request packets. Our contribution is twofold: first, we provide a detailed analysis of how routers are most commonly configured to respond to expiring packets; next, we show that for the vast majority of routers, the measured round-trip time is not affected by the probing rate.

The Software-Defined Networking approach permits to realize new policies. In OpenFlow in particular, a controller decides on behalf of the switches which forwarding rules must be installed and where. However with this flexibility comes the challenge of the computation of a rule allocation matrix meeting both high-level policies and the network constraints such as memory or link capacity limitations. Nevertheless, in many situations (e.g., data-center networks), the exact path followed by packets does not severely impact performances as long as packets are delivered according to the endpoint policy. It is thus possible to deviate part of the traffic to alternative paths so to better use network resources without violating the endpoint policy. In this paper, we propose a linear optimization model of the rule allocation problem in resource constrained OpenFlow networks with relaxing routing policy. We show that the general problem is NP-hard and propose a polynomial time heuristic, called OFFICER, which aims to maximize the amount of carried traffic in under-provisioned networks. Our numerical evaluation on four different topologies shows that exploiting various paths allows to increase the amount of traffic supported by the network without significantly increasing the path length.

Data offloading from the cellular network to low-cost WiFi has been the subject of several research works in the last years. In-network caching has also been studied as an efficient means to further reduce cellular network traffic. In this paper we consider a scenario where mobile users can download popular contents (e.g., maps of a city, shopping information, social media, etc.) from WiFi-enabled caches deployed in an urban area. We study the optimal distribution of contents among the caches (i.e., what contents to put in each cache) to minimize users' access cost in the whole network. We argue that this optimal distribution does not necessarily provide geographic fairness, i.e., users at different locations can experience highly variable performance. In order to mitigate this problem, we propose two different cache coordination algorithms based on gossiping. These algorithms achieve geographic fairness while preserving the minimum access cost for end users.

Information-centric networking (ICN) leverages content demand redundancy and proposes in-network caching to reduce network and servers load and to improve quality of experience. In this paper, we study the interaction between in-network caching of ICN and Additive Increase Multiplicative Decrease (AIMD) end-to-end congestion control with a focus on how bandwidth is shared, as a function of content popularity and cache provisioning. As caching shortens AIMD feedback loop, the download rate of AIMD is impacted. Supported by an analytical model based on Discriminatory Processor Sharing and real experiments, we observe that popular contents benefit from caching and realize a shorter download time at the expense of unpopular contents, which see their download time inflated by a factor bounded by 1/1-ρ where ρ is the network load. This bias can be removed by redefining congestion control to be delay independent or by over-provisioning link capacity at the edge so that to compensate for the greediness of popular contents.

Emerging challenged networks require new protocols and strategies to cope with a high degree of mobility, high delays and unknown, possibly non-existing routes within the network. Researchers have proposed different store-carry-and-forward protocols for data delivery in challenged networks. These have been complemented with appropriate drop and scheduling policies that deal with the limitations of the nodes' buffers and the limited duration of opportunistic encounters in these networks. Nevertheless, the vast majority of these protocols and strategies are designed for end-to-end transmissions. Yet, a paradigm shift from the traditional way of addressing the endpoints in the network has been occurring towards content-centric networking. To this end, we present CEDO, a content-centric dissemination algorithm for challenged networks. CEDO aims at maximizing the total delivery-rate of distributed content in a setting where a range of contents of different popularity may be requested and stored, but nodes have limited resources. It achieves this by maintaining a delivery-rate utility per content that is proportional to the content miss rate and that is used by the nodes to make appropriate drop and scheduling decisions. This delivery-rate utility can be estimated locally by each node using unbiased estimators fed by sampled information on the mobile network obtained by gossiping. Both simulations and theory suggest that CEDO achieves its set goal, and outperforms a baseline LRU-based policy by 72%, even in relatively small scenarios. The framework followed by CEDO is general enough to be applied to other global performance objectives as well.

The identification of Internet traffic applications is very important for ISPs and network administrators to protect their resources from unwanted traffic and prioritize some major applications. Statistical methods are preferred to port-based ones since they don't rely on the port number, which can change dynamically, and to deep packet inspection since they also work for encrypted traffic. These methods combine the statistical analysis of the application packet flow parameters, such as packet size and inter-packet time, with machine learning techniques. Other successful approaches rely on the way the hosts communicate and their traffic patterns to identify applications. In this paper, we propose a new online method for traffic classification that combines the statistical and host-based approaches in order to construct a robust and precise method for early Internet traffic identification. Without loss of generality we use the packet size as the main feature for the classification and we benefit from the traffic profile of the host (i.e., which application and how much) to refine the classification and decide in favor of this or that application. The host profile is then updated online based on the result of the classification of previous flows originated by or addressed to the same host. We evaluate our method on real traces using several applications. The results show that leveraging the traffic pattern of the host ameliorates the performance of statistical methods. They also prove the capacity of our solution to derive profiles for the traffic of Internet hosts and to identify the services they provide.

In order to achieve data delivery in Delay Tolerant Networks (DTN), researchers have proposed the use of store-carry-and-forward protocols: a node there may store a message in its buffer and carry it along for long periods of time, until an appropriate forwarding opportunity arises. This way, messages can traverse disconnected parts of the network. Multiple message replicas are often propagated to further increase delivery probability. This combination of long-term storage and message replication imposes a high storage and bandwidth overhead. Thus, efficient scheduling and drop policies are necessary to 1) decide on the order by which messages should be replicated when contact durations are limited, and 2) which messages should be discarded when nodes' buffers operate close to their capacity. In this paper, we propose a practical and efficient joint scheduling and drop policy that can optimize different performance metrics, such as average delay and delivery probability. We first use the theory of encounter-based message dissemination to derive the optimal policy based on global knowledge about the network. Then, we introduce a method that estimates all necessary parameters using locally collected statistics. Based on this, we derive a distributed scheduling and drop policy that can approximate the performance of the optimal policy in practice. Using simulations based on synthetic and real mobility traces, we show that our optimal policy and its distributed variant outperform existing resource allocation schemes for DTNs. Finally, we study how sampled statistics can reduce the signaling overhead of our algorithm and examine its behavior under different congestion regimes. Our results suggest that close to optimal performance can be achieved even when nodes sample a small percentage of the available statistics.

Video streaming represents a large fraction of Internet traffic. Surprisingly, little is known about the network characteristics of this traffic. In this paper, we study the network characteristics of the two most popular video streaming services, Netflix and YouTube. We show that the streaming strategies vary with the type of the application (Web browser or native mobile application), and the type of container (Silverlight, Flash, or HTML5) used for video streaming. In particular, we identify three different streaming strategies that produce traffic patterns from non-ack clocked ON-OFF cycles to bulk TCP transfer. We then present an analytical model to study the potential impact of these streaming strategies on the aggregate traffic and make recommendations accordingly.

The identification of Internet applications is important for ISPs and network administrators to protect the network from unwanted traffic and prioritize some major applications. Statistical methods are widely used since they allow to classify applications according to their statistical signatures. They combine the statistical analysis of flow parameters, such as packet size and inter-packet time, with machine learning techniques. Previous works are mainly based on the packet size and the directions of the packets. In this work we make a complete study about the inter-packet time to prove that it is also a valuable information for the classification of Internet traffic. We discuss how to isolate the noise due to the network conditions and extract the time generated by the application. We present a model to preprocess the inter-packet time and use the result as input to the learning process. We discuss an iterative approach for the on line identification of the applications and we evaluate our method on two different real traces. The results show that the inter-packet time is an important parameter to classify Internet traffic.

The remarkable growth of the Internet infrastructure and the increasing heterogeneity of applications and users' behavior make more complex the manageability and monitoring of ISP networks and raises the cost of any new deployment. The main consequence of this trend is an inherent disagreement between existing monitoring solutions and the increasing needs of management applications. In this context, we present the design of an adaptive centralized architecture that provides visibility over the entire network through a network-wide cognitive monitoring system. Practically, given a measurement task and a constraint on the volume of collected information, the proposed architecture drives the sampling rates on the interfaces of network routers to achieve the maximum possible accuracy, while adapting itself to any change in network traffic conditions. We illustrate our work with an accounting application whose purpose is to estimate the volume of aggregate flows across a backbone transit network. The paper provides a global study of the functioning of the proposed system and the impact of the different parameters on its behavior. The performance of our system is validated in typical scenarios over an experimental platform we developed for the purpose of the study.

Packet sampling can greatly reduce traffic measurement overhead in high-speed networks. At the same time, this operation introduces estimation errors that have to be carefully handled to ensure a reasonable measurement accuracy. Recently, a frequency-based approach has been proposed to catch the impact of such errors in bitrate estimation of IP flows at various time-scales. In particular, a closed-form expression for the signal-to-noise ratio has been derived as a function of the packet sampling probability, the bin size, and some basic information about the flow (i.e., first and second order moments of the packet size, and long term average packet-rate). In this work, we adopt this model to design a real-time algorithm that properly tunes the temporal resolution of bitrate estimates available at a router, with the goal of assuring a target measurement accuracy. The effectiveness of the proposed approach has been demonstrated by computer simulations, carried out using real packet traces.

Ubiquity of portable computing devices coupled with wide availability of wireless communication present new impor- tant opportunities for applications involving media-rich content dissemination. However, as access networks become increasingly more heterogeneous, seamless data delivery across internets consisting of a variety of network technology becomes a real challenge. In this demonstration, we showcase a system that enables content dissemination over heterogeneous internets consisting of wired, infrastructure-based and infrastructure-less wireless networks that may be prone to intermittent connectivity. Using an efficient, yet flexible buffer management scheme, we are able to address application-specific performance requirements such as average delay, delivery probability, energy efficiency, etc. Our system uses the Message Delivery in Heterogeneous, Disruption-prone Networks (MeDeHa) [2]) framework to deliver messages across a heterogeneous internet coupled with History-Based Scheduling and Drop (HBSD) buffer management [1] as a way to optimize resources provided by opportunistic networks. MeDeHa, which is described in detail in [2], provides seamless data delivery over interconnecting networks of different types, i.e., infrastructure-based and infrastructure-less networks. MeDeHa's comprehensive approach to bridging infrastructure-based and infrastructureless networks also copes with intermittent connectivity. For this demonstration, we showcase a "complete stack" solution featuring, from to top to bottom, the DTN2 "bundle" layer, HBSD as an "external router" to DTN2, and MeDeHa, which handles message delivery. We have implemented, on a Linux-based testbed, (i) the MeDeHa framework, (ii) the HBSD [3] external router for the DTN2 [4] architecture.

In network measurement systems, packet sampling techniques are usually adopted to reduce the overall amount of data to collect and process. Being based on a subset of packets, they introduce estimation errors that have to be properly counteracted by using a fine tuning of the sampling strategy and sophisticated inversion methods. This problem has been deeply investigated in the literature with particular attention to the statistical properties of packet sampling and to the recovery of the original network measurements. Herein, we propose a novel approach to predict the energy of the sampling error in the real time estimation of traffic bitrate, based on spectral analysis in the frequency domain. We start by demonstrating that the error introduced by packet sampling can be modeled as an aliasing effect in the frequency domain. Then, we derive closed-form expressions for the Signal-to-Noise Ratio (SNR) to predict the distortion of traffic bitrate estimates over time. The accuracy of the proposed SNR metric is validated by means of real packet traces. Furthermore, a comparison with respect to an analogous SNR expression derived using classic stochastic tools is proposed, showing that the frequency domain approach grants for a higher accuracy when traffic rate measurements are carried out at fine time granularity.

In network measurement systems, packet sampling techniques are usually adopted to reduce the overall amount of data to collect and process. Being based on a subset of packets, they hence introduce estimation errors that have to be properly counteracted by a fine tuning of the sampling strategy and sophisticated inversion methods. This problem has been deeply investigated in the literature with particular attention to the statistical properties of packet sampling and the recovery of the original network measurements. Herein, we propose a novel approach to predict the energy of the sampling error on the real time traffic volume estimation, based on a spectral analysis in the frequency domain. We start by demonstrating that errors due to packet sampling can be modeled as an aliasing effect in the frequency domain. Then, we exploit this theoretical finding to derive closed-form expressions for the Signal-to-Noise Ratio (SNR), able to predict the distortion of traffic volume estimates over time. The accuracy of the proposed SNR metric is validated by means of real packet traces.

It is of utmost importance for the network research community to have access to tools and testbeds to explore future directions for Internet traffic monitoring and engineering. Although many experimental solutions exist today, they tend to be highly specialized or to have a limited availability and openness. Through this work, we outline the monitoring capabilities limitations of these facilities and we present our emulation platform for network wide traffic monitoring as an answer to these limitations. Our platform presents a new approach for the emulation of Internet traffic and for its monitoring across the different routers. Through our solution, we put at the disposal of users a real traffic emulation service coupled to a set of libraries and tools capable of Cisco NetFlow data export and collection, the overall destined to run advanced applications for network wide traffic monitoring and optimization.

Swarming peer-to-peer systems play an increasingly instrumental role in Internet content distribution. It is therefore important to better understand how these systems behave in practice. Recent research efforts have looked at various protocol parameters and have measured how they affect system performance and robustness. However, the importance of the strategy based on which peers establish connections has been largely overlooked. This work utilizes extensive simulations to examine the default overlay construction strategy in BitTorrent systems. Based on the results, we identify a critical parameter, the maximum allowable number of outgoing connections at each peer, and evaluate its impact on the robustness of the generated overlay. We find that there is no single optimal value for this parameter using the default strategy. We then propose an alternative strategy that allows certain new peer connection requests to replace existing connections. Further experiments with the new strategy demonstrate that it outperforms the default one for all considered metrics by creating an overlay more robust to churn. Additionally, our proposed strategy exhibits optimal behavior for a well-defined value of the maximum number of outgoing connections, thereby removing the need to set this parameter in an ad-hoc manner.

We address the issue of asserting the accuracy of coordinates advertised by nodes of Internet coordinate systems during distance estimations. Indeed, some nodes may lie deliberately about their coordinates to mount various attacks against applications and overlays. Our proposed method consists in two steps: 1) establish the correctness of a node's claimed coordinate (which leverages our previous work on securing the coordinates embedding phase using a Surveyor infrastructure); and 2) issue a time limited validity certificate for each verified coordinate. Validity periods are computed based on an analysis of coordinate inter-shift times observed on PlanetLab, and shown to follow a long-tail distribution (lognormal distribution in most cases, or Weibull distribution otherwise). The effectiveness of the coordinate certification method is validated by measuring the impact of a variety of attacks on distance estimates.

Packet sampling techniques introduce measurement errors that should be carefully handled in order to correctly characterize the network behavior. In the literature several works have studied the statistical properties of packet sampling and the way it should be inverted to recover the original network measurements. Here we take the new direction of studying the spectral properties of packet sampled traffic. A novel technique to model the impact of packet sampling is proposed based on a theoretical analysis of network traffic in the frequency domain. Moreover, a real-time algorithm is also presented to detect the spectrum portion of the network traffic that can be restored once packet sampling has been applied. Preliminary experimental results are reported to validate the proposed approach.

As a result of the wide spread of new efficient mobile communication devices, users are more and more willing to find on these devices (PDAs, Smart-phones, etc) applications similar to the ones they are using on their desktop computers. We go one step in this direction and propose in this demo a software we have designed and implemented to enable content sharing among spontaneous communities of mobile users using wireless multihop connections. Our contribution consists in designing, implementing and testing a tracker-less Bit-Torrent like application adapted to wireless ad-hoc networks. The proposed architecture includes two principal components: a membership management service and a content sharing service. As classical tracker-based Bit-Torrent membership management and peer discovery are unfeasible in ad hoc networks, we design the membership management service as a distributed tracker overlay that connects peers involved in the same sharing session. Using the membership information provided by the tracker overlay, the content sharing service schedules the data transfer connections among the session members by leveraging the multihop routing feature of wireless ad-hoc networks. Our test-bed composed of PDAs and smart-phones equipped with WIFI adapters and Windows Mobile 6 operating system is used to validate the feasibility of our application and to evaluate its performance.

Delay Tolerant Networks are wireless networks where disconnections may occur frequently due to propagation phenomena, node mobility, and power outages. Propagation delays may also be long due to the operational environment (e.g. deep space, underwater). In order to achieve data delivery in such challenging networking environments, researchers have proposed the use of store-carry-and-forward protocols: there, a node may store a message in its buffer and carry it along for long periods of time, until an appropriate forwarding opportunity arises. Additionally, multiple message replicas are often propagated to increase delivery probability. This combination of long-term storage and replication imposes a high storage overhead on untethered nodes (e.g. handhelds). Thus, efficient buffer management policies are necessary to decide which messages should be discarded, when node buffers are operated close to their capacity. In this paper, we propose efficient buffer management policies for delay tolerant networks. We show that traditional buffer management policies like drop-tail or drop-front fail to consider all relevant information in this context and are, thus, sub-optimal. Using the theory of encounter-based message dissemination, we propose an optimal buffer management policy based on global knowledge about the network. Our policy can be tuned either to minimize the average delivery delay or to maximize the average delivery rate. Finally, we introduce a distributed algorithm that uses statistical learning to approximate the global knowledge required by the the optimal algorithm, in practice. Using simulations based on a synthetic mobility model and real mobility traces, we show that our buffer management policy based on statistical learning successfully approximates the performance of the optimal policy in all considered scenarios. At the same time, our policy outperforms existing ones in terms of both average delivery rate and delivery delay.

Delay tolerant networks (DTN) are wireless networks where disconnections may occur frequently. In order to achieve data delivery in DTNs, researchers have proposed the use of store-carry-and-forward protocols: there, a node may store a message in its buffer and carry it along for long periods of time, until an appropriate forwarding opportunity arises. Multiple message replicas are often propagated to increase delivery probability. This combination of long-term storage and replication imposes a high storage and bandwidth overhead. Thus, efficient scheduling and drop policies are necessary to: (i) decide on the order by which messages should be replicated when contact durations are limited, and (ii) which messages should be discarded when nodespsila buffers operate close to their capacity. In this paper, we propose an efficient joint scheduling and drop policy that can optimize different performance metrics, like average delay and delivery probability. Using the theory of encounter-based message dissemination, we first propose an optimal policy based on global knowledge about the network. Then, we introduce a distributed algorithm that can approximate the performance of the optimal algorithm, in practice. Using simulations based on a synthetic mobility model and a real mobility trace, we show that our optimal policy and its distributed variant outperform existing resource allocation schemes for DTNs, such as the RAPID protocol [4], both in terms of average delivery ratio and delivery delay.

This paper addresses the issue of the security of Internet Coordinate Systems,by proposing a general method for malicious behavior detection during coordinate computations. We first show that the dynamics of a node, in a coordinate system without abnormal or malicious behavior, can be modeled by a Linear State Space model and tracked by a Kalman filter. Then we show, that the obtained model can be generalized in the sense that the parameters of a filtercalibrated at a node can be used effectively to model and predict the dynamic behavior at another node, as long as the two nodes are not too far apart in the network. This leads to the proposal of a Surveyor infrastructure: Surveyor nodes are trusted, honest nodes that use each other exclusively to position themselves in the coordinate space, and are therefore immune to malicious behavior in the system.During their own coordinate embedding, other nodes can thenuse the filter parameters of a nearby Surveyor as a representation of normal, clean system behavior to detect and filter out abnormal or malicious activity. A combination of simulations and PlanetLab experiments are used to demonstrate the validity, generality, and effectiveness of the proposed approach for two representative coordinate embedding systems, namely Vivaldi and NPS.

Modeling TCP is fundamental for understanding Internet behavior. The reason is that TCP is responsible for carrying a huge quota of the Internet traffic. During last decade many analytical models have attempted to capture dynamics and steady-state behavior of standard TCP congestion control algorithms. In particular, models proposed in literature have been mainly focused on finding relationships among the throughput achieved by a TCP flow, the segment loss probability, and the round trip time (RTT) of the connection, which the flow goes through. Recently, Westwood+ TCP algorithm has been proposed to improve the performance of classic New Reno TCP, especially over paths characterized by high bandwidth-delay products. In this paper, we develop an analytic model for the throughput achieved by Westwood+ TCP congestion control algorithm when in the presence of paths with time-varying RTT. The proposed model has been validated by using the ns-2 simulator and Internet-like scenarios. Validation results have shown that this model provides relative prediction errors smaller than 10%. Moreover, it has been shown that a similar accuracy is achieved by analogous models proposed for New Reno TCP.

Confronted with the generalization of monitoring in operational networks, researchers have proposed placement algorithms that can help ISPs deploy their monitoring infrastructure in a cost effective way, while maximizing the benefits of their infrastructure. However, a static placement of monitors cannot be optimal given the short-term and long-term variations in traffic due to re-routing events, anomalies and the normal network evolution. In addition, most ISPs already deploy router embedded monitoring functionalities. Despite some limitations (inherent to being part of a router), these monitoring tools give greater visibility on the network traffic but raise the question on how to configure a network-wide monitoring infrastructure that may contain hundreds of monitoring points.We reformulate the placement problem as follows. Given a network where all links can be monitored, which monitors should be activated and which sampling rate should be set on these monitors in order to achieve a given measurement task with high accuracy and low resource consumption? We provide a formulation of the problem, an optimal algorithm to solve it, and we study its performance on a real backbone network.

Devices in disruption tolerant networks (DTNs) must be able to communicate robustly in the face of short and infrequent connection opportunities. Unfortunately, one of the most inexpensive, energy-efficient and widely deployed peer-to-peer capable radios, Bluetooth, is not well-suited for use in a DTN. Bluetooth's half-duplex process of neighbor discovery can take tens of seconds to complete between two mutually undiscovered radios. This delay can be larger than the time that mobile nodes can be expected to remain in range, resulting in a missed opportunity and lower overall performance in a DTN. This paper proposes a simple, cost effective, and high performance modification to mobile nodes to dramatically reduce this delay: the addition of a second Bluetooth radio. We showed through analysis and simulation that this dual radio technique improves both connection frequency and duration. Moreover, despite powering two radios simultaneously, nodes using dual radios are more energy efficient, spending less energy on average per second of data transfered.

The end-to-end delay among a set of peers can be easily and scalably estimated using a landmark approach Peers' coordinates are deduced from delay measurements to a small number N of landmarks L{L1,..., Ln}. We wonder whether it is feasible to use such distributed solution for bandwidth estimation. This requires that each peer determines its bandwidth vector by measuring the direct and reverse bandwidth on its path with each landmark. The bandwidth vectors of two peers are then used to estimate the bandwidth between them in both ways. For a couple of peers, we denote by (i) direct path the network path that joins them directly using IP routing, and by (ii) indirect path the path that joins them via a landmark node. N indirect paths (N being the number of landmarks) are assigned to each direct path. Our idea is to estimate the bandwidth of the direct path using those of the indirect paths. The indirect paths that have more links common with the direct path, are assigned more weight in the estimation function. We consider different estimation functions and we study the impact of the landmarks' locations on the accuracy of the estimations.

In this paper, an enhanced scalable proximity model is described. This work investigates the bandwidth estimation problem and it is impact on the quality of service of the file sharing application. The file transfer application over the TCP protocol is considered and it can be encountered in the emerging file sharing P2P applications or in the replicated Web server context. Since the impact of the bandwidth estimation is the main concern, the large TCP transfer due to its sensitivity to the parameter is considered. Thus considering the delay alone for proximity characterization is far from being optimal for large file transfer applications

Network coding is a new transmission paradigm that proved its strength in optimizing the usage of network resources. In this paper, we evaluate the gain from using network coding for file sharing applications running on top of wireless mesh networks. With extensive simulations carried out on a simulator we developed specifically for this study, we confirm that network coding can improve the performance of the file sharing application, but not as in wired networks. The main reason is that nodes over wireless cannot listen to different neighbors simultaneously. Nevertheless, one can get more from network coding if the information transmission is made more diverse inside the network. We support this argument by varying the loss rate over wireless links and adding more sources.

The replication of digital content in overlay networks makes the identification of the best server an interesting problem. In this paper, our aim is to reduce the content transfer time, which we define as the time required to download a digital content by a client using TCP. Our scheme consists of ranking the servers from the best one to the worst one based on a metric that corresponds to a prediction of the content transfer time from each server. Our prediction function considers the critical performance parameters that have an impact on the quality of the transfer, such as the load of the servers and the characteristics of the path between the client and the servers. Once the servers are ranked, the client can download the content on point-to-point from the best server, or in parallel from a subset of servers at the top of the list (best servers). Our experimental results show that our approach for identifying the best server(s) outperforms the existing classical solutions. Moreover, these results show that the proposed metric predicts the transfer time of a content with around 96% accuracy.

Most of the theoretical work on sampling has addressed the inversion of general traffic properties such as flow size distribution, average flow size, or total number of flows. In this paper, we make a step towards understanding the impact of packet sampling on individual flow properties. We study how to detect and rank the largest flows on a link. To this end, we develop an analytical model that we validate on real traces from two networks. First we study a blind ranking method where only the number of sampled packets from each flow is known. Then, we propose a new method, protocol-aware ranking, where we make use of the packet sequence number (when available in transport header) to infer the number of non-sampled packets from a flow, and hence to improve the ranking. Surprisingly, our analytical and experimental results indicate that a high sampling rate (10% and even more depending on the number of top flows to be ranked) is required for a correct blind ranking of the largest flows. The sampling rate can be reduced by an order of magnitude if one just aims at detecting these flows or by using the protocol-aware method.

We present an analytical model that accounts for the positions of stations with respect to the access point (AP) while evaluating the performance of the 802.11 MAC layer. Our work is based on the Bianchi model where the performance of the 802.11 MAC layer is computed using a discrete time Markov chain, but where all stations are implicitly assumed to be located at the same distance from the AP. In our model, given the position of one station, we compute its saturation throughput while considering the positions of other concurrent stations. Further, our model provides the total saturation throughput of the medium. We solve the model numerically and we show that the saturation throughput per station is strongly dependent not only on the station's position, but also on the positions of the other stations. Results confirm that a station achieves a higher throughput when it is closer to the AP, but bring out that there is a distance threshold above which the throughput decrease is fast and significant. When a station is far from the AP compared to the other stations, it contends for the bandwidth not used by the other stations. We believe that our model is a good tool to dimension 802.11 wireless access networks and to study their capacities and their performances.

In this paper, we present a model for TCP/IP congestion control mechanism. The rate at which data is transmitted increases linearly in time until a packet loss is detected. At this point, the transmission rate is divided by a constant factor. Losses are generated by some exogenous random process which is assumed to be stationary ergodic. This allows us to account for any correlation and any distribution of inter-loss times. We obtain an explicit expression for the throughput of a TCP connection and bounds on the throughput when there is a limit on the window size. In addition, we study the effect of the Timeout mechanism on the throughput. A set of experiments is conducted over the real Internet and a comparison is provided with other models that make simple assumptions on the inter-loss time process. The comparison shows that our model approximates well the throughput of TCP for many distributions of inter-loss times.

We study in this paper the performance of TCP over a wireless link implementing hybrid FEC/ARQ-SR at the link layer. The study is done by simulating a large number of TCP transfers over a wireless link showing Bernoulli errors. We are motivated by how to tune link-level error recovery, e.g. amount of FEC, persistency of ARQ, so as to minimize the latency of TCP. We provide results for different physical characteristics of the wireless link (delay, error rate), different traffic loads and different file sizes. Our main finding is that the latency of TCP always improves with the persistency of ARQ, except for some extreme cases where the delay is large, files are small, and the loss rate is low. When adding FEC, the latency of TCP improves then deteriorates, and the deterioration is more pronounced in the case of large files, high loss rate and small delay. Another finding of our study is that with the hybrid mechanism, the wireless link is able to carry more traffic than when FEC and ARQ-SR are separately used.

The emerging widespread use of real-time multimedia applications over wireless networks makes the support of quality of service (QoS) a key problem. In this paper, we focus on QoS support mechanisms for IEEE 802.11 wireless ad-hoc networks. First, we review limitations of the upcoming IEEE 802.11e enhanced DCF (EDCF) and other enhanced MAC schemes that have been proposed to support QoS for 802.11 ad-hoc networks. Then, we describe a new scheme called adaptive fair EDCF that extends EDCF, by increasing the contention window during deferring periods when the channel is busy, and by using an adaptive fast backoff mechanism when the channel is idle. Our scheme computes an adaptive backoff threshold for each priority level by taking into account the channel load. The new scheme significantly improves the quality of multimedia applications. Moreover, it increases the overall throughput obtained both in medium and high load cases. Simulution results show that our new scheme outperforms EDCF and other enhanced schemes. Finally, we show that the adaptive fair EDCF scheme achieves a high degree of fairness among applications of the same priority level.

The throughput of AIMD protocols in general and of TCP in particular, has been computed in many existing works by modeling the round-trip time as a constant and thus replacing the round-trip time by its expectation. There are however many scenarios in which the delays of packets vary, causing a variation of the round-trip time. Many typical scenarios occur in wireless and mobile networks. We propose in this paper an analytical model that accounts for the variability of delay, while computing the throughput of an AIMD protocol. We derive a closed-form expression for the throughput that illustrates the impact of the variability of delay. We show by analysis and simulation, that an increase in the variability of delay improves the performance of an AIMD protocol. Thus, an analytical model that only considers the average delay could underestimate the performance of an AIMD protocol in scenarios where delay is variable.

We study an approach to quality-of-service (QoS) that offers end-users the choice between two service classes defined according to their level of transmission protection. The fully protected (FP) class offers end-users a guarantee of survivability in the case of a single-link failure; all FP traffic is protected using a 1:1 protection scheme at the wavelength-division multiplexing (WDM) layer. The best effort protected (BEP) class is not protected; instead restoration at the IP layer is provided. The FP service class mimics what Internet users receive today. The BEP traffic is designed to run over the large amounts of unused bandwidth that exist in today's Internet. The goal is to increase the load carried on backbone networks without reducing the QoS received by existing customers. To support two such services, we have to solve two problems: the off-line problem of mapping logical links to pairs of disjoint fiber paths, and an on-line scheduling problem for differentiating packets from two classes at the IP layer. We provide an algorithm based on a Tabu Search meta-heuristic to solve the mapping problem, and a simple but efficient scheduler based on weighted fair queueing for service differentiation at the IP layer. We consider numerous requirements that carriers face and illustrate the tradeoffs they induce. We demonstrate that we can successfully increase the total network load by a factor between three and ten and still meet all the carrier requirements.

The IEEE 802.11 medium access control (MAC) protocol provides a contention-based distributed channel access mechanism for mobile stations to share the wireless medium, which may introduce a lot of collisions in case of overloaded active stations. Slow contention window (CW) decrease scheme is a simple and efficient solution for this problem. In this paper, we use an analytical model to compare the slow CW decrease scheme to the IEEE 802.11 MAC protocol. Several parameters are investigated such as the number of stations, the initial CW size, the decrease factor value, the maximum backoff stage and the coexistence with the RequestToSend and ClearToSend (RTS/CTS) mechanism. The results show that the slow CW decrease scheme can efficiently improve the throughput of IEEE 802.11, and that the throughput gain is higher when the decrease factor is larger. Moreover, the initial CW size and maximum backoff stage also affect the performance of slow CW decrease scheme.

We propose two novel on-line estimation algorithms to determine the size of a dynamic multicast group. We first use a Wiener filter to derive an optimal estimator for the membership size of the session in case the join process is Poisson and the lifetime of participants is distributed exponentially. We next develop the best first-order linear filter from which we derive an estimator that holds for any lifetime distribution. We apply this approach to the case where the lifetime distribution is hyperexponential. Both estimators hold under any traffic regime. Applying both estimators on real traces corresponding to video sessions, we find that both schemes behave well, one of which performs slightly better than the other in some cases. We further provide guidelines on how to tune the parameters involved in both schemes in order to achieve high quality estimation while simultaneously avoiding feedback implosion.

Our goal is to design a traffic model for uncongested IP backbone links that is simple enough to be used in network operation, and that is protocol and application agnostic in order to be as general as possible. The proposed solution is to model the traffic at the flow level by a Poisson shot-noise process. In our model, a flow is a generic notion that must be able to capture the characteristics of any kind of data stream. We analyze the accuracy of the model with real traffic traces collected on the Sprint IP backbone network. Despite its simplicity, our model provides a good approximation of the real traffic observed in the backbone and of its variation. Finally, we discuss three applications of our model to network design and management.

We present an analytical model for the calculation of network load and drop probabilities in a TCP/IP network with general topology. First we formulate our model as a nonlinear complementarity problem. Then we transform the model into two equivalent formulations: fixed point formulation and nonlinear programming formulation. These equivalent formulations provide efficient computational procedures for the solution of our model. Furthermore, with the help of the fixed point formulation we are able to prove the existence of a solution. Our model has the main advantage of not requiring the pre-definition of bottleneck links. The model also takes into account the receiver congestion window limitation. Our approach can be used for TCP/IP networks with drop tail buggers as well as for TCP/IP networks with active queue management buggers. We solve the problem for some network examples and we show how the distribution of load varies with network parameters. The distribution of load is sometimes counter-intuitive which cannot be detected by other models making prior assumptions on the locations of bottlenecks.

LCA

We study in this paper a TCP-like linear-increase multiplicative-decrease flow control mechanism. We consider congestion signals that arrive in batches according to a Poisson process. We focus on the case when the transmission rate cannot exceed a certain maximum value. We write the Kolmogorov equations and we use Laplace transforms to calculate the distribution of the transmission rate in the steady state as well as its moments. Our model is particularly useful to study the behavior of TCP, the congestion control mechanism in the Internet. By a simple transformation, the problem can be reformulated in terms of an equivalent M/G/1 queue, where the transmission rate in the original model corresponds to the workload in the 'dual' queue. The service times in the queueing model are not i.i.d., and they depend on the workload in the system.

In interactive voice applications, FEC schemes are necessary for the recovery from packet losses. These schemes need to be simple with a light coding and decoding overhead in order to not impact the interactivity. The objective of this paper is to study a well known simple FEC scheme, in which for every packet n, some redundant information is added in some subsequent packet n+/spl phi/. If packet n is lost, it will be reconstructed in case packet n+/spl phi/ is well received. The quality of the reconstructed copy of packet n will depend on the amount of information on packet n we add to packet n+/spl phi/. We propose a detailed queueing analysis based on a ballot theorem and obtain simple expressions for the audio quality as a function of the amount of redundancy and its relative position to the original information. The analysis shows that this FEC scheme does not scale well and that the quality will deteriorate for any amount of FEC and for any offset /spl phi/.

We present a technique for identifying repetitive information transfers and use it to analyze the redundancy of network traffic. Our insight is that dynamic content, streaming media and other traffic that is not caught by today's Web caches is nonetheless likely to derive from similar information. We have therefore adapted similarity detection techniques to the problem of designing a system to eliminate redundant transfers. We identify repeated byte ranges between packets to avoid retransmitting the redundant data.

No abstract available.

No abstract available.