Internet Exchange Points facilitate traffic exchange through Route Servers that use BGP for path selection without considering latency. This paper presents LARS, the first system to enable latency-aware path optimization at IXP route servers. It requires no changes to BGP and no bilateral coordination among members, thus directly enabling latency-aware routing for all IXP members. LARS continuously probes the route-server RIB, annotates routes with RTT-derived BGP communities, and enables members to (i) enforce a configurable latency radius that suppresses high-latency routes and (ii) prefer lower-latency alternatives when multiple paths exist. We evaluate LARS at a very large IXP using a production-level prototype that demonstrates stable latency estimation, effective tail-latency control via selective redistribution, and potential to optimize BGP best paths.

Distributed Denial of Service (DDoS) attacks are among the most critical cybersecurity threats, jeopardizing the stability of even the largest networks and services. The existing range of mitigation services predominantly filters at the edge of the Internet, thus creating unnecessary burden for network infrastructures. Consequently, we present IXP Scrubber, a Machine Learning (ML) based system for detecting and filtering DDoS traffic at the core of the Internet at Internet Exchange Points (IXPs) which see large volumes and varieties of DDoS. IXP Scrubber continuously learns DDoS traffic properties from neighboring Autonomous Systems (ASes). It utilizes BGP signals to drop traffic for certain routes (blackholing) to sample DDoS and can thus learn new attack vectors without the operator's intervention and on unprecedented amounts of training data. We present three major contributions: i) a method to semi-automatically generate arbitrarily large amounts of labeled DDoS training data from IXPs' sampled packet traces, ii) the novel, controllable, locally explainable and highly precise two-step IXP Scrubber ML model, and iii) an evaluation of the IXP Scrubber ML model, including its temporal and geographical drift, based on data from 5 IXPs covering a time span of up to two years.

Amplification Distributed Denial of Service (DDoS) attacks' traffic and harm are at an all-time high. To defend against such attacks, distributed attack mitigation platforms, such as traffic scrubbing centers that operate in peering locations, e.g., Internet Exchange Points (IXP), have been deployed in the Internet over the years. These attack mitigation platforms apply sophisticated techniques to detect attacks and drop attack traffic locally, thus, act as sensors of attacks. However, it has not yet been systematically evaluated and reported to what extent coordination of these views by different platforms can lead to more effective mitigation of amplification DDoS attacks. In this paper, we ask the question: "Is it possible to mitigate more amplification attacks and drop more attack traffic when distributed attack mitigation platforms collaborate?"

Due to the COVID-19 pandemic, many governments imposed lock-downs that forced hundreds of millions of citizens to stay at home. The implementation of confinement measures increased Internet traffic demands of residential users, in particular, for remote working, entertainment, commerce, and education, which, as a result, caused traffic shifts in the Internet core.

Booter services continue to provide popular DDoS-as-a-service platforms and enable anyone irrespective of their technical ability, to execute DDoS attacks with devastating impact. Since booters are a serious threat to Internet operations and can cause significant financial and reputational damage, they also draw the attention of law enforcement agencies and related counter activities. In this paper, we investigate booter-based DDoS attacks in the wild and the impact of an FBI takedown targeting 15 booter websites in December 2018 from the perspective of a major IXP and two ISPs. We study and compare attack properties of multiple booter services by launching Gbps-level attacks against our own infrastructure. To understand spatial and temporal trends of the DDoS traffic originating from booters we scrutinize 5 months, worth of inter-domain traffic. We observe that the takedown only leads to a temporary reduction in attack traffic. Additionally, one booter was found to quickly continue operation by using a new domain for its website.

Network attacks, including Distributed Denial-of-Service (DDoS), continuously increase in terms of bandwidth along with damage (recent attacks exceed 1.7 Tbps) and have a devastating impact on the targeted companies/governments. Over the years, mitigation techniques, ranging from blackholing to policy-based filtering at routers, and on to traffic scrubbing, have been added to the network operator's toolbox. Even though these mitigation techniques provide some protection, they either yield severe collateral damage, e.g., dropping legitimate traffic (blackholing), are cost-intensive, or do not scale well for Tbps level attacks (ACL filtering, traffic scrubbing), or require cooperation and sharing of resources (Flowspec).

Multi-layer resilience is one of the prominent new concepts for modern carrier networks; it efficiently combines the advantages of the optical and the packet layer. However, not all features offered by modern optical transport networks can be fully used by the packet layer yet. In case of a fiber cut, an optical protection mechanism restores the original IP topology after a short transient time. But such an optical restoration is expected to use a new, longer light-path, which in turn might affect the optical capacity of the link. Bit-rate flexible optical transceivers are able to utilize the remaining optical capacity efficiently by adapting the network link capacity accordingly. However, the packet layer is not able to cope with fluctuating link capacities; often the policy is to rather shut down the link completely instead of using the remaining capacity. Consequently, this paper proposes a Segment Routing-based approach with superimposed Software-defined Networking (SDN) to allow the IP network to benefit from these new features. The minimally invasive, gradual deployment of the system is investigated, while keeping other proven and resilient technologies and systems unmodified. Using the topology and traffic matrix of a large German carrier the feasibility of such a deployment is evaluated.

Reputation networks are an important building block of distributed systems whenever reliability of nodes is an issue. However, reputation ratings can easily be undercut: colluding nodes can spread good ratings for each other while third parties are hardly able to detect the fraud. There is strong analytical evidence that reputation networks cannot be constructed in a way to guarantee security. Consequently, only statistical approaches are promising. This work pursues a statistical approach inspired by the idea that colluding node's behavior changes the local structure of a reputation network. To measure these structural changes, we extend a graph analysis method originating from molecular biology and combine it with a machine learning approach to analyze fingerprints of node's interactions. We evaluate our method using an adaptive Peer-to-Peer (P2P) streaming system and show that a correct classification of up to 98% is possible.

Video traffic is the main driver of Internet traffic volume. Thus, content providers and Content Delivery Networks (CDNs) are searching for ways to provide reliable video transmission at a low cost. Hybrid CDN/Peer-to-Peer (P2P) deployments like Akamai NetSession have been shown to combine the high reliability of a CDN backbone and the low cost of P2P networks. In the near future, the biggest barrier for user adoption will fall: the installation of a dedicated P2P client software will be replaced by website embedded browser-to-browser communication logic. However, this requires the explicit consent of users, and, since users need to share their upload capacity, their willingness to participate in such a system. In this work, the efficiency of incentive mechanisms trading a higher Quality of Experience (QoE) of video transmission for user's consent to utilize their upload capacity are investigated. This is the first study to investigate the question of incentives in distributed, adaptive streaming systems from a user perspective using a crowd working approach. The work presents results from 192 test subjects. We identify three classes of users and show how behavioral economics can be utilized to increase the impact of an incentive scheme.

Adaptive streaming systems gain rising relevance for streaming services. Therefore, the same video is offered in multiple quality versions to clients for adaptation during playback. However, optimizing adaptation in a Quality of Experience (QoE) centric way is difficult. Current systems maximize bit rate, ignoring that different types of adaptation (resolution, framerate, quantization) correlate differently and in a non-linear way with user's perception. User validated video quality metrics can provide precise quality information. However, measurements of state-of-the-art metrics show either high computational intensity or weak correlation with subjective tests. This makes large-scale offline quality assessment processing intensive while real-time constrained scenarios like live streaming and video conferencing are hardly supportable. Consequently, this work presents the Real-Time Video Quality Metric (RT-VQM), a real-time, Graphics Processing Unit (GPU) supported version of the widely used Video Quality Metric (VQM). RT-VQM introduces efficient filtering operations, hardware-supported scaling and high-performance feature pooling. The approach outperforms VQM by a factor of 30, thus enabling a real-time assessment of up to 9 parallel video stream representations up to High Definition (HD) 720 resolution at 30fps.

Energy efficiency of networking equipment has become an important factor for private consumers as well as business customers in the recent years. This trend is driven by two factors: the proliferation of new mobile devices as well as energy prices as cost driver. Battery capacity as well as dynamic distribution of workloads adapted to energy costs are an active field of research. This demo contributes to this discussion by presenting a framework allowing the assessment of energy costs along the content distribution chain.

Live streaming of large-scale events such as the Olympic Games with a huge number of viewers is challenging, as the streaming infrastructure needs to scale fast and big, and often in an unpredictable manner. Peer-to-peer (P2P) live streaming (Peercasting) has proven to be beneficial in these scenarios, as resources are scaling inherently with the number of nodes. However, churn behavior in a node's neighborhood may result in fluctuating downstream bandwidth and thus freezing (stalling) playback. Related work tries to mitigate this effect by using layered video codecs, focusing on single-dimensional scalability in mesh-pull based systems. Yet, the benefits of multidimensional scalability (resolution, frame rate, and quantization) combined with coexisting pull-/push mechanisms introduced by modern hybrid P2P streaming architectures have not been studied in detail. Consequently, this work proposes a new scheduling algorithm taking these aspects into account. The evaluation shows large benefits for end-users by reducing the frequency of stalls by 90% even under extreme conditions.

Mobile data consumption has seen a considerable rise and is predicted to further increase. This is mainly caused by mobile video consumption. A promising solution is offloading cellular traffic to WiFi. Here, Multipath TCP (MPTCP) enables a seamless user experience for handover and load balancing. However, its influence on the energy consumption of smartphones is not yet well analyzed. To this end, this paper determines the energy cost of using MPTCP for mobile constant bit rate streaming on two Nexus devices and compares it to the power consumption using single interface TCP streaming. From this, recommendations for an optimal assignment of sub-streams to interfaces are derived. The measurements indicate that using MPTCP on certain smartphones causes a 20% lower energy expense compared to the cost of both interfaces individually. However, MPTCP using multiple interfaces should only be used if the requested data rate cannot be supported by a single interface.

Video streaming constitutes the dominant portion of today’s traffic on the Internet and will grow in the coming years. In order to provide for a low cost distribution of bulky video content, Peer-to-Peer (P2P) approaches are a viable way to cut down server bandwidth cost by utilizing user’s upstream bandwidth to redistribute data. However, users need an incentive to participate in such a system. The related work on incentive schemes has focused on using bandwidth contribution as a measure for contribution to the system’s performance, ignoring that user perceived Quality of Experience (QoE) is not necessarily maximized by maximizing bandwidth, but by delivering the right data in the right order. Consequently, this work presents QTrade, a topology agnostic incentive scheme for adaptive P2P video streaming systems based on user-validated video quality metrics. QTrade is evaluated on top of an existing adaptive streaming overlay. The results show that QTrade utilizes bandwidth more efficiently by providing incentive to distribute parts of the video with a high QoE. Moreover, up to 70% less and shorter rebuffering events are observed for cooperative peers while maintaining a 10 to 11 times worse performance in terms of rebuffering events for non-cooperative peers.

A wireless Ad Hoc network consisting of a source and multiple receiving nodes is considered. The source wants to transmit a common message throughout the whole network. The message has to be spread in a multi-hop fashion, as the transmit powers at the source and the nodes are limited. The goal of this paper is to find the multi-hop broadcast tree with a minimum energy consumption in the network. To reach this goal, a new decentralized game theoretic approach is proposed which considers the following two aspects jointly for the first time: Firstly, it optimizes the transmit powers at the source and at the individual intermediate nodes. Secondly, it employs maximum ratio combining at the receiving nodes following the fact that a node can receive several copies of the message from different sources in different time slots. The game is modeled such that the nodes are incentivized to forward the message to their neighbors. In terms of the total transmit energy, the results show that the proposed algorithm outperforms other conventional algorithms.

Computation offloading is an upcoming approach to increase battery life of mobile devices overburdened by resource-consuming applications. In multi-hop networks, computation offloading poses new challenges since intermediate devices are required to relay tasks of others along the path to the server. The decision of a device about whether to offload or not depends thus on the provided energy of relay devices and on the decisions of other offloading devices since relay resources need to be shared. This also implies that for energy minimization, optimal decisions are topology-dependent. This paper introduces a novel theoretical framework for energy minimization of computation offloading in multi-hop wireless networks which formulates the energy minimization problem as a binary linear problem. Proving its equivalence to a multi-dimensional knapsack problem allows us to specify a greedy heuristic, which shows very good performance, with a maximal deviation of less than 5% from the optimal results. From simulations and analytical results for different topologies, we derive under which conditions computation offloading in multi-hop networks is beneficial.

For transmitting data in scenarios showing a high user density, infrastructure based and multihop Ad hoc communication can be combined to benefit from the reliability of a stable backbone network and the increased coverage of multihop communication. Such scenarios have been investigated from a cross layer perspective in the recent years mainly focusing on pure performance optimization. However, the question of providing incentives to nodes to forward data has largely been ignored in the cross layer domain, even though providing incentives is vital for the network: each node represents a user comparing his or her satisfaction and the cost to decide on his or her participation. A likely reason for the gap in cross layer incentive research is the necessity to model users as well as the network in order to express a user's utility, which requires knowledge in both fields. In order to foster future research in the area of cross layer incentive schemes, this work proposes a general cross layer simulation model combining user and network models. Moreover, an instantiation of the simulation model for the use case of live video broadcasting is presented.

The collaboration of Internet service providers (ISPs) and content distribution network (CDN) providers was shown to be beneficial for both parties in a number of recent works. Influencing CDN edge server (surrogate) selection allows the ISP to manage the rising amount of traffic emanating from CDNs to reduce the operational expenditures (OPEX) of his infrastructure, e.g., by preventing peered traffic. At the same time, including the ISP's hidden network knowledge in the surrogate selection process influences the quality of service a CDN provider can deliver positively. As a large amount of CDN traffic is video-on-demand traffic, this paper investigates the topic of CDN/ISP collaboration from a perspective of high-volume long-living flows. These types of flows are hardly manageable with state-of-the-art Dynamic Name Service (DNS)-based redirection, as a reassignment of flows during the session is difficult to achieve. Consequently, varying load of surrogates caused by flash crowds and congestion events in the ISP's network are hard to compensate. This paper presents a novel approach promoting ISP and CDN collaboration based on a minimal deployment of software-defined networking switches in the ISP's network. The approach complements standard DNS-based redirection by allowing for a migration of high-volume flows between surrogates in the backend even if the communication has state information, such as Hyper Text Transfer Protocol sessions. In addition to a proof-of-concept, the evaluation identifies factors influencing performance and shows large performance increases when compared to standard DNS-based redirection.

Peer-to-Peer (P2P) has proven to be a scalable approach for content distribution, while reducing the load for the content provider. As P2P systems depend on each node's participation, reciprocal incentive mechanisms for stimulating contribution are a major building block in those systems. However, reciprocal incentive schemes fall short for resource-poor devices, as they lead to service degradation for those nodes. This paper presents vINCENT, an incentive scheme allowing resource-poor devices to seek support from resource-rich devices. The novel scheme is based on virtual nodes which form a trusted domain around multiple devices. vINCENT has been implemented and evaluated in a P2P live streaming scenario. The solution provides good playback quality to all devices within a virtual node, given a sufficient contribution of the virtual node as a whole. At the same time, free riding nodes are isolated efficiently and the game theoretic properties of reciprocity are preserved.

The transmission of video content accounts for a large share of today's Internet traffic. While Video-on-Demand (VoD) substantially contributes to this, live streaming events such as video broadcasts from the Olympic Games can cause very high traffic volumes in the short term as well. Such peaks along with high fluctuations triggered by sudden changes in the behavior of users make the design of live streaming systems particularly challenging. Peer-to-Peer (P2P) has proven to be a scalable approach for disseminating content to a large number of users. Accordingly, the body of research offers numerous P2P live streaming approaches tailored towards specific scenarios and assumptions. However, no single approach is able to perform well under all possible conditions. Keeping up a high performance when conditions are changing is a challenge, since topology management and scheduling mechanisms cannot be exchanged easily. Therefore, this paper proposes TRANSIT, a new approach going beyond existing works in that it supports transitions between different live streaming mechanisms. TRANSIT makes different configurations of such mechanisms seamlessly exchangeable to enable the optimal choice of configurations for a wide range of live streaming scenarios. The approach is evaluated using measurement- and trace-based workloads. The results show that TRANSIT is able to maintain a high performance at a low overhead in highly fluctuating environments, whereas static configurations show serious performance degradations.

The Peer-to-Peer (P2P) paradigm offers scalable means to perform bulk data distribution, e.g., for small businesses which cannot afford huge upfront investments, by incorporating user's resources in the dissemination process. Due to the proliferation of smartphones with wireless broadband connectivity and the increasing convergence of fixed and mobile platforms, a growing number of users are expected to participate in P2P content distribution networks wirelessly. However, the P2P approach only works if users are willing to contribute resources. A commonly applied incentive scheme is the well-known Tit-for-Tat approach, where each peer is forced to contribute as much bandwidth to the network as he consumes. Nevertheless, reciprocal schemes discriminate resource poor mobile devices in terms of energy and upload bandwidth, as they are device-bound instead of being user-bound. In this work, an incentive scheme featuring virtual nodes is presented, which allows mobile devices to seek help from other devices owned by its user, e.g., the user's home gateway or a supporting cloud instance. Preliminary results are presented in the scope of a P2P streaming scenario.

The proliferation of wireless broadband technologies and mobile devices has led to an increase in mobile traffic, especially due to a rapid growth of real-time entertainment such as video streaming on smartphones. Mobile peer-to-peer-based content distribution schemes can help to relieve infrastructure based mobile networks, but require participating nodes to provide resources which can drain their battery. Thus, the goal is to exploit mobile peers' resources while minimizing and balancing the energy consumption over all participating devices. Simulation models considering energy consumption lack precision, because they abstract away important parts of the hardware. On the other hand, prototypical energy measurements are more precise, but require a time consuming implementation and assessment. To this end, this demo paper presents a mobile P2P video streaming and benchmarking platform which enables to assess and compare the energy consumption of different approaches in a precise manner through live assessments at runtime. The demonstrated platform includes a simple, yet high-performance tree-based mobile P2P streaming overlay which can be utilized to easily implement and assess further streaming overlay approaches.