direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

Past talks, student talks, and other events of 2011

On modeling Internet topology and traffic for studying malicious software mitigation and other network traffic related activities
Speaker:
Enric Pujol, Fraunhofer FOKUS
Type:
Project group meeting (Projektgruppentreffen, PGT)
Time:
13 December 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
Many network operations involve monitoring and analyzing network traffic data. Working with such data is an active area that presents many challenges. In this talk, I will present my work in this area. This includes the work reported on my diploma thesis and my subsequent research activities in the framework of a few European, national and industrial projects. The talk will be organized in three parts.
In the first part, I will present my activities on intrusion detection and mitigation. This includes the study of the epidemic dynamics of self-propagating malicious software and involves modeling network topologies and traffic. In this part, I will also present some initial work on containment methods that are based on regulating traffic at the flow level and are intended to operate on local name servers.
In the second part, I will present my activities on network traffic monitoring for network management purposes. This involves analyzing data passively captured on the Gn and Iu-PS interfaces of mobile networks. Specifically, I will talk about an approach that is based on filtering and biased flow and packet sampling to isolate network flows that are interesting for the operators of this type of networks.
In the third part, I will present my activities on performing active traffic measurements to validate Service Level Agreements (SLAs) of real time traffic applications. This involves monitoring network-level performance metrics, such as the one-way delay to detect degradations in the quality of such applications. In more detail, I will talk about an approach to enhance Cisco's UDP jitter operation.
Approximations for virtual networks embeddings
Speaker:
Carlo Fürst, TUB student
Type:
Diplom thesis final talk
Time:
06 December 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
In the last years, there has been much research on network virtualization. Different perspectives on different usecases led to a broad variety of approaches to embed virtual networks on substrates. Outstanding due to its high complexity is the datacenter virtualization usecase, where an arbitrary virtual network request has to be mapped to a substrate in a way that satisfies all constraints. Current techniques to generate this embedding have either unwanted technical limitations or unacceptable runtimes for a real world usecases. This thesis establishes a way to reduce the runtime, without introducing unwanted limitations, by sacrificing optimality.
On the Cost of Concurrency in Transactional Memory
Speaker:
Srivatsan Ravi
Type:
Project group meeting (Projektgruppentreffen, PGT)
Time:
29 November 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
The promise of software transactional memory (STM) is to combine an easy-to-use programming interface with an efficient utilization of the concurrent-computing abilities provided by modern machines. But does this combination come with an inherent cost?
We evaluate the cost of concurrency by measuring the amount of expensive synchronization that must be employed in an STM implementation that ensures positive concurrency, i.e., allows for concurrent transaction processing in some executions. We focus on two popular progress conditions that provide positive concurrency: progressiveness and permissiveness.
We show that in permissive STMs, providing a very high degree of concurrency, a transaction may perform a linear number of expensive synchronization patterns with respect to its read-set size. In contrast, progressive STMs provide a very small degree of concurrency but, as we demonstrate, can be implemented using at most one expensive synchronization pattern per transaction. However, we show that even in progressive STMs, a transaction has to “protect”' (e.g., by using locks or strong synchronization primitives) a linear amount of data with respect to its write-set size. Our results suggest that achieving high degrees of concurrency in STM implementations may bring a considerable synchronization cost.
Economic Aspects of VNet Embeddings
Speaker:
Arne Ludwig
Type:
Diplom thesis final talk
Time:
22 November 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
We introduce the Price of specification (PoS) as a measure for the effective value of a VNet for the Physical Infrastructure Provider w.r.t. the accuracy of the VNet specification. This measure may help a PIP to adjust his pricing structure to improve his market position. We derive a formula to calculate the expected number of embeddings in specific scenarios depending on the detail level of the incoming VNets specification, which can then be used to generate the PoS. Furthermore we study simulations to get insights on the different impacts on the PoS. We distinguish between the Data center and the outsourcing scenario. The focus in the outsourcing scenario lies on the geographical location of the VNets as they have additional placement constraints. The different parameters we study can be divided into those effecting only the substrate, those effecting only the VNets and those effecting both.
Quantitative Analysis of Physical Layer and Link Layer Measurements in WiFi Networks
Speaker:
Julius Schulz-Zander, TU Berlin Diplom student
Type:
Diplom thesis final talk
Time:
10 November 2011
16:00–17:00
Place:
TEL 1118/19
Abstract:
In this thesis, we present a measurement study on the quantification of lost transmission opportunities in IEEE 802.11 networks, which is an increasingly more important problem due to a higher number of IEEE 802.11-based deployments. Lost transmission opportunities can arise from both unwanted traffic and link impairments. We perform passive measurements to gain information about unwanted traffic which lead to a reduction of available airtime and active measurements to gain information on link impairments which lead to packet loss or unfairness. In both cases, we follow a cross layer approach, since measuring lost transmission opportunities depends on information from two layers: (1) the physical (PHY) layer which provides signal quality and transmission statistics and (2) the medium access-control (MAC) layer which provides IEEE 802.11 protocol specific information, and throughput and loss statistics. We present our measurement methodology on how to collect network wide measurements with off-the shelf WiFi hardware. Using our measurement framework, we collect several network wide measurements that contain, for example, the hardware states of the WiFi card, link statistics and packet traces. From passive measurements, we show that a significant overhead of IEEE 802.11 control and management traffic exist, reducing the wireless network capacity. In addition, we show that, with active measurements a close estimation of the root-cause of loss due to link impairments is possible.
Social Engineering Approach to Computer Systems Design
Speaker:
Pan "Ben" Hui
Type:
Project group meeting (Projektgruppentreffen, PGT)
Time:
8 November 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
Many computer networks have dual properties. They are physical networks and at the same time human networks. It is tremendously important to exploit this social networking feature to design more efficient and more reliable communication systems. In this talk, I will illustrate, with several examples, how social networks can be integrated into system design. I will start by introducing several human mobility experiments with diversified scenarios and presenting two important social metrics, community and centrality, commonly observed in the experiments. Next, I will talk about how these social metrics can be applied to forwarding in opportunistic networks, mobile data offloading for cellular networks, and design of spam protection systems.
Augmenting the mobile experience through code offload
Speaker:
Eduardo Cuervo, Duke University
Type:
Talk
Time:
27 October 2011
11:00–12:00
Place:
Auditorium 2
Abstract:
One of the mobile industry biggest challenges consists in overcoming the resource limitations in mobile devices. The most important of this limitations is battery capacity, being suggested by technology trends that this limitation is here to stay. Similarly, mobile devices are unable to perform computationally intensive tasks such as face or speech recognition, natural language processing, high-end graphics or augmented reality. My research focuses on enabling new types of mobile interactions and applications by hiding this limitations from both users and developers through code offload. In this talk I will describe a system we built in this direction and then I will discuss my ongoing research on the same problem. I will present MAUI (short for Mobile Assistance Using Infrastructure), a system that enables fine-grained energy-aware offload of mobile code to nearby server infrastructure. MAUI uses the benefits of a managed code environment to offer the best of both worlds: it supports fine-grained code offload to maximize energy savings with minimal burden on the programmer. MAUI decides at run-time which methods should be remotely executed, driven by an optimization engine that achieves the best energy savings possible under the mobile device’s current connectivity constrains.
Bio:
Eduardo Cuervo is a 6th year PhD candidate in Computer Science at Duke University. His research interests include mobile and pervasive computing, networks and distributed systems. Cuervo has a BS in Computer Science from the Monterrey Institute of Technology and Higher Studies in Mexico City.
Practical Online Admission Control with Deterministic Performance Guarantees
Speaker:
Xin Li, TU Berlin master student
Type:
Master's thesis final talk
Time:
25 October 2011
14:00–15:00
Place:
New Room: TEL 6
Abstract:
The compressed digital video is one of the most prevalent type of traffic in integrated services networks. Due to its high degree of burstiness, a common belief is that it is hard to maintain a high network utilization and also provide quantitative performance guarantees on desirable video properties such as bounded delay. In this thesis we propose and evaluate an online admission control scheme for video flows, whose key characteristic stands in a very low storage and computational complexity. In particular, we determine the minimum amount of memory to store video traffic, based on which we can analytically determine whether incoming flows can be admitted or not subject to some predefined worst-case delay constraints. Our study relies on the framework of the deterministic network calculus, which has evolved as an attractive methodology for deriving worst-case performance guarantees in data networks.
A New Paradigm for the Internet. “What not Where”.
An overview of the CCNx protocol (Part II)
Speaker:
Aidan Walton
Type:
Project group meeting (Projektgruppentreffen, PGT)
Time:
18 October 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
The presentation discusses the drivers behind Content Centric Networking in general. Comparing and contrasting its functional elements with that of TCP/IP. CCNx architectural principles are highlighted. Further discussion looks in more detail at the CCNx protocol; packet types and their semantics and descriptions of a wire level CCNx exchange. The CCN Node model is described, explaining the different data structures used and how they operate in the current prototype implementation. The various research areas are presented, highlighting the most significant problems that need to be addressed with an overview of some of the current thinking. Finally the presentation will cover some preliminary baseline testing of the protocol and some comparisons with TCP.
Season: A Dynamic Load Balancer for Virtual Environments
Speakers:
Thomas Krenc, Tobias Jacobowitz, Sebastian Garn
(TU Berlin master students)
Type:
Master's project final talk
Time:
11 October 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
Currently deployed client-server architectures for Massively Multiplayer Online Games (MMOG) serve millions of concurrent clients. Scaling MMOGs to this magnitude is challenging. Several techniques have been proposed in order to distribute clients load on multiple servers, e.g. spatial decomposition of the virtual world, or 'sharding', to name some. Most of the proposed solutions rely on static assignments of users to servers in the virtual world, thus, they can not cope with the volatile behavior of clients. As a consequence, some servers are overloaded, while others are under-loaded. For MMOGs, computation power is a scarce resource that has to be optimized in order to offer better quality of experience to the end users as well as improve the reliability of the system.
In this work, we argue that it is possible to dynamically assign users to servers in order to equalize the load of the servers allocated in a MMOG. We propose a distributed load-balancing algorithm that dynamically assign clients to servers based on the current activity in the system. Our distributed algorithm also allows an MMOG to re-assign the area of responsibility between servers in a distributed, scalable, and robust way. To evaluate the performance of our algorithm we build a prototype, called Season, that can serve as a building block for a MMOG. Our results, with real and synthetic data, show that the load among servers in an MMOG is well balanced, while the number of re-assignments of users to servers is minimal when Season is used.
Efficiency of Wireless Networks
Speaker:
Magnús M. Halldórsson, Reykjavik University
Type:
Summer School
Time:
22 September 2011
13:00–20:00
Place:
TEL 1118/19
Abstract:
What is the capacity of wireless networks and how can they be best utilized? Such questions form the foundation of the MAC layer in wireless networks.
We examine recent work on maximizing the throughput of arbitrary wireless networks in the physical model. The focus is on efficient and practical algorithms, preferably distributed, and on the intuition that follows from their analysis.
Ressources:
Talk slides (PDF, 2,3 MB)
Misleading Stars: What Cannot Be Measured In The Internet?
Speaker:
Gilles Trédan
Type:
Project group meeting (Projektgruppentreffen, PGT)
Time:
06 September 2011
15:00–16:00
Place:
Auditorum 1
Abstract:
Traceroute measurements are one of our main instruments to shed light onto the structure and properties of today's complex networks such as the Internet. This paper studies the feasibility and infeasibility of inferring the network topology given traceroute data from a worst-case perspective, i.e., without any probabilistic assumptions on, e.g., the nodes' degree distribution. We attend to a scenario where some of the routers are anonymous, and propose two fundamental axioms that model two basic assumptions on the traceroute data: (1) each trace corresponds to a real path in the network, and (2) the routing paths are at most a factor 1/α off the shortest paths, for some parameter α ∈ (0,1]. In contrast to existing literature that focuses on the cardinality of the set of (often only minimal) inferrable topologies, we argue that a large number of possible topologies alone is often unproblematic, as long as the networks have a similar structure. We hence seek to characterize the set of topologies inferred with our axioms. We introduce the notion of star graphs whose colorings capture the differences among inferred topologies; it also allows us to construct inferred topologies explicitly. We find that in general, inferrable topologies can differ significantly in many important aspects, such as the nodes' distances or the number of triangles. These negative results are complemented by a discussion of a scenario where the trace set is best possible, i.e., "complete". It turns out that while some properties such as the node degrees are still hard to measure, a complete trace set can help to determine global properties such as the connectivity.
DES-SERT, a Framework for implementing reactive layer 2.5 routing protocols
Speaker:
Philipp Schmidt, FU Berlin
Type:
Project group meeting (Projektgruppentreffen, PGT)
Time:
29 August 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
Implementing reactive routing protocols for wireless mesh networks  on commodity operating systems is hard. Once one leaves the traditional  architecture of having a routing table containing nexthops for all reachable destinations, he has the choice of either hacking the architecture to somehow implement the protocol or implementing most aspects of the routing on his own. Usually, this leads to almost unmaintainable code and also raises the question whether the results of a comparison between protocols are just implementation specific or protocol inherent.
In this talk I will introduce you to DES-SERT,  a simple and extensible layer 2.5 routing framework for testbeds, which has been implemented in 2009 to aid implementation and evaluation of reactive routing protocols in the DES-Testbed at Freie Universität Berlin to overcome the issues stated above.
Philipp Schmidt's personal website
A New Paradigm for the Internet. “What not Where”.
An overview of the CCNx protocol
Speaker:
Aidan Walton
Type:
Project group meeting (Projektgruppentreffen, PGT)
Time:
23 August 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
The presentation discusses the drivers behind Content Centric Networking in general. Comparing and contrasting its functional elements with that of TCP/IP. CCNx architectural principles are highlighted. Further discussion looks in more detail at the CCNx protocol; packet types and their semantics and descriptions of a wire level CCNx exchange. The CCN Node model is described, explaining the different data structures used and how they operate in the current prototype implementation. The various research areas are presented, highlighting the most significant problems that need to be addressed with an overview of some of the current thinking. Finally the presentation will cover some preliminary baseline testing of the protocol and some comparisons with TCP.
Bowlmap 2.0: A network visualization tool
Speaker:
Francisco Herrera-Luque and Benjamin Vahl
Type:
Internship final talk
Time:
09 Aug 2011
15:00–16:00
Place:
Auditorium 1
Abstract:
In this talk, we will present the new version of our network visualization tool, bowlmap 2.0. The new version of our map tool allows more flexibility in terms of different types of measurements that can be supported. Additionally, it is more efficient in terms of sending to client browsers. In the talk, we will briefly discuss the features of the old version of the map and discuss the necessary changes to the architecture in order to support higher flexibility and low overhead. The new bowl architecture will be discussed in detail and guidelines for how to display any type of measurements will be presented. We believe bowlmap 2.0 will serve as an effective tool for network monitoring and debugging for various different projects within INET.
Towards Improved Control and Troubleshooting for Operational Networks
Speaker:
Andreas Wundsam
Type:
PhD Thesis Defense
Time:
15 July 2011
14:00–16:00
Place:
Auditorium 3
(TEL 20)
Abstract:
Over the past decade, operational networks, have grown tremendously in size, performance and importance. We expect this trend to continue as more and more services traditionally provided by the local computer move to the cloud, e.g., file storage services and office applications.

In spite of this, our ability to control and manage these networks remains painfully inadequate, and our visibility into the network limited. Controlling and troubleshooting operational networks is challenging due to a multitude of reasons, including their intrinsic properties (distribution, scale, geographic spread), but also due to lacking architectural support and an insufficient tool set.

In this thesis, we explore how to improve our control over networks and our abilities to debug and troubleshoot problems. Due to the extreme diversity of the environments, we do not strive for a one-size-fits-all solution, but propose and evaluate several approaches tailored to specific important scenarios and environments. We emphasize network centric approaches that can be implemented locally and are transparent to the end hosts.

In the talk, we highlight two approaches: We propose Mirror VNets as a primitive that enables safer evolution and improved debugging abilities for complex network services. To this end, a production VNet is paired with a Mirror VNet in identical state and configuration.

Also, we explore how Software Defined Network architectures, e.g., OpenFlow, can be leveraged to enable record and replay troubleshooting for Networks. We propose and evaluate OFRewind, the first system that
enables practical record and replay in operational networks, even in the presence of black-box devices than cannot be modified or instrumented. We present several case studies that underline its utility. Our
evaluation shows that OFRewind scales at least as well as current controller implementations and does not significantly impact the scalability of an OpenFlow controller domain.
Oblivious Collaboration
Speaker:
Eli Gafni, UCLA
Type:

Talk
Time:
30 June 2011
11:00–12:00
Place:
Sputnik
(TEL 17)
Abstract:
Communication is a crucial ingredient in every kind of collaborative work. But what is the least possible amount of communication required for a given task? We formalize this question by introducing a new framework for distributed computation, called oblivious protocols. We investigate the power of this model by considering two concrete examples, the musical chairs task MC(n,m) and the well-known Renaming problem. The MC(n,m) game is played by n players (processors) with m chairs. Players can occupy chairs, and the game terminates as soon as each player occupies a unique chair. Thus we say that player P is in conflict if some other player Q is occupying the same chair, i.e., termination means there are no conflicts. By known results from distributed computing, if m ≤ 2n−2, no strategy of the players can guarantee termination. However, there is a protocol with m = 2n−1 chairs that always terminates. Here we consider an oblivious protocol where in every time step the only communication is this: an adversarial scheduler chooses an arbitrary nonempty set of players, and for each of them provides only one bit of information, specifying whether the player is currently in conflict or not. A player notified not to be in conflict halts and never changes its chair, whereas a player notified to be in conflict changes its chair according to its deterministic program. Remarkably, even with this minimal communication termination can be guaranteed with only m = 2n−1 chairs. Likewise, we obtain an oblivious protocol for the Renaming problem whose name-space is small as that of the optimal nonoblivious distributed protocol. Other aspects suggest themselves, such as the efficiency (program length) of our protocols. We make substantial progress here as well, though many interesting questions remain open.

Joint work with: Yehuda Afek, Yakov Babichenko, Uriel Feige, Nati Linial, Benny Sudakov
Impact of Location on Content Delivery
Speaker:
Bernhard Ager
Type:

PhD Thesis Defense

Time:
25 June 2011
17:30–19:30
Place:
Auditorium 3
A First Look at a Commercial Hybrid Content Delivery System
Speaker:
Bruce Maggs, Duke University and Akamai Technologies
Type:

Talk

Time:
24 June 2011
14:00–15:00
Place:
Auditorium 1
(TEL 20)
Abstract:
This talk describes a commercial hybrid content delivery service, Akamai's Download Manager (DLM).  The DLM extends Akamai's fixed-infrastructure network of edge servers by employing peer-to-peer elements.  The DLM currently includes tens of millions of peers in over 230 countries. Using a one-month trace from the production system, we study the characteristics of the peer population, the mobility of peers between different networks and locations, the pattern of download activity, and the performance of peer-assisted versus fixed-infrastructure downloads.

Joint work with Ming-Chen Zhao, Paarijat Aditya, Yin Lin, Andreas Harberlen, Peter Druschel, and William Wishon.

Bio:
see website

Distributed Infrastructures for Online Social Networks
Speaker:
Doris Schiöberg
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
21 June 2011
15:00–16:30
Place:
Auditorium 1
New room: FR 5028, Franklinstraße
Abstract:
Online Social Networks (OSNs) are immensely popular today and store a lot of personal data. But there are some issues: What happens to this data when the OSN shuts down? Is the data lost? Can a user get full control (including deletion) over his uploaded content?
An alternative to server-based OSNs are distributed approaches, such as PeerSoN or Diaspora. They enable easily scalable redundancy and privacy control, without needing to rely on a single operating entity. Yet the question is whether distributed OSNs are feasible, i.e., can they provide all the convenient features that, e.g., Facebook offers. This talk focuses on the problem of understanding the availability of user content within such a distributed OSN.
OF-Rewind – Enabling troubleshooting for operational networks
Speaker:
Andreas Wundsam
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
7 June 2011
15:00–17:00
Place:
Auditorium 1
Abstract:
Debugging operational networks can be a daunting task, due to their size, distributed state, and the presence of black box components such as commercial routers and switches, which are poorly instrumentable and only coarsely configurable. The debugging tool set available to administrators is limited, and provides only aggregated statistics (SNMP), sampled data (NetFlow/sFlow), or local measurements on single hosts (tcpdump). We leverage split forwarding architectures such as OpenFlow to add record and replay debugging capabilities to networks – a powerful, yet currently lacking approach. We present the design of OFRewind, which enables scalable, multi-granularity, coherent recording and coordinated replay in a network, with fine-grained, dynamic, centrally orchestrated control over both record and replay. Thus, OFRewind helps operators to reproduce software errors, identify data-path limitations, or configuration errors.

Understanding Cross-Layer Effects on Quality of Experience for Video over NGMN
Speaker:
Amir Mehmood
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
31 May 2011
15:45–16:30
Place:
Auditorium 1
Abstract:
The evolution of wireless network standards, e.g., GSM/GPRS, UMTS, WiFi, WiMAX, and end-user devices has paved the way towards Next Generation Mobile Networks (NGMN), where users are always connected through multiple radio access networks. NGMN technologies target to improve the user experience especially for mobile data and multimedia services, which are in line with user expectations evident from, for instance, the increasingly popular mobile web video streaming. To understand the quality that can be offered to the user, we compare the Quality of Experience (QoE) for web streaming in a prototype NGMN testbed with WiFi and 3G UMTS/HSDPA support. We use CUBIC TCP as the transport protocol as it is typically the default TCP variant, e.g., in Android phones. We complement the QoE estimations with network Quality of Service (QoS) parameters such as throughput and delay, and transport layer statistics. The results of our evaluation show that (i) video QoE remains stable in WiFi even with high packet loss, (ii) QoE in HSDPA is sensitive to packet loss even for low loss rates due to high variations in the network QoS, namely, throughput and delay, (iii) the decrease in QoE and QoS in HSDPA is due to its negative interactions with the aggressive congestion control of CUBIC TCP, and (iv) handover from WiFi to HSDPA degrades QoE.
On Revealing the ARQ Mechanism of MSTV
Speaker:
Oliver Hohlfeld
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
31 May 2011
15:00–15:45
Place:
Auditorium 1
Abstract:
Ensuring a high customer satisfaction by monitoring Quality of Experience (QoE) aspects has become common practice for service providers. Such monitoring solutions, together with underlying QoE models, are mostly limited to measures captured in the core or access network and may thus neglect the QoE impact of recovery mechanisms deployed at client-side, e.g., FEC or ARQ. This limitation makes QoE models prone to mispredict QoE and consequently may lead the operators to misleading interpretations of customer experience. In this paper, we empirically study the behavior of the Microsoft TV Set-top Box (STB) with respect to the deployed ARQ recovery mechanism. As the ARQ implementation details of the STB are proprietary, we implement and simulate three ARQ algorithms of different complexities and evaluate their performance by comparing with corresponding empirical measurements. This comparison reveals insights into the ARQ scheme implemented in the STB. Moreover, it leads us to speculate that MSTV uses simple ARQ schemes which are sufficient to drastically improve the QoE in the presence of a multitude of loss patterns.
Deep diving into performance and scalability of OpenFlow implementations
Speaker:
Nadi Sarrar

Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
17 May 2011
ca. 17:15–18:00
Place:
Auditorium 1
Abstract:
Oflops, an open and generic tool that enables the rapid development of use-case tests for OpenFlow-enabled switches will be presented. Oflops combines advanced hardware instrumentation, for accuracy and performance, with a flexible and extensible software framework. Oflops supports pre-defined test scenarios that measure the capabilities and bottlenecks between the forwarding engine of the switch and the remote control-application.
Alongside a demonstration of performance and comparison of OpenFlow switches, we observe their strengths and weaknesses both in their flow updating as well as monitoring capabilities. These observations are crucial to understand the applicability of OpenFlow in the context of specific use-cases. Furthermore, given the current lack of switch-support to provide information about when OpenFlow commands have been completed, we illustrate the importance of a tool such as Oflops that observes control plane behavior from the data plane.

Revisiting Content Distribution in the Smartphone Age
Speaker:
Benjamin Frank
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
17 May 2011
ca. 16:30–17:15
Place:
Auditorium 1
Abstract:
There is a phenomenal growth in traffic volume in 3G networks after the introduction of smart phones and tablet computers. Recent studies showed that more than 50% of this traffic is served by Content Delivery Networks, such as Akamai or Google. Despite the fact that the 3G network architecture differs from wired ones, the way content is delivered in such networks is the same. Thus, it does not only inherit the limitations of traditional content delivery, but also has unique properties of its own, e.g., modem power states or the wireless link, that influences content delivery. Yet no study has quantified the implications these unique properties of 3G networks have on content delivery. To take this into account, we will present an enhanced measurement methodology for content delivery in 3G networks from the one introduced in our IMC2010 submission "Improving Content Delivery Using Provider-aided Distance Information". Furthermore we discuss our research agenda which aims at identifying 3G network specific differences and quantifying their impact on content delivery in 3G networks as well as proposing possible ways for improvement.
Virtual Networks
Speaker:
Gregor Schaffrath
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
17 May 2011
ca. 15:45–16:30
Place:
Auditorium 1
Abstract:
tba
Getting to know better the Internet through games and topology techniques
Speaker:
Eirini Spartinou
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
10 May 2011
16:00–16:45
Place:
Auditorium
Abstract:
The problem of predicting links in networks can be placed under the general taxonomy of Link mining tasks. It is a rather new area of research interest that first arose from Social Networks. A recent method based on the notion of hierarchical organization is tackling the link forecasting problem. This method achieves that in a way that suggests hierarchy could be used to reveal important information on complex networks.
802.11 interference measurements
Speaker:
Thomas Hühn
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
3 May 2011
17:00–18:00
Place:
Auditorium 1
Abstract:
Ressource allocation in 802.11 multi-hop networks focuses mostly on rate-control rather than transmission power and carrier sensing. The parameter space of adjusting those factors is hugh and testing all combinations in a separat experiment is just unfeasible. In this talk we start with the process and method of meaningful reduction of this parameter space. Some new metrics like 802.11 mac states and the implemented tools to measure those metrics are shown. We present results about the impact and interaction that certain factors of the ressource allocation process have on different metrics like noise and wireless radio states.
Practical Online Admission Control with Deterministic Performance Gurantees
Speaker:
Xin Li
Type:

Master's thesis introductory talk

Time:
3 May 2011
16:00–17:00
Place:
Auditorium 1
Abstract:
The compressed digital video is one of the most prevalent type of traffic in integrated services networks. Due to its high degree of burstiness, a common belief is that it is hard to maintain a high network utilization and also to provide quantitative performance guarantees on desirable video properties such as bounded delay. In this thesis we propose and evaluate an online admission control scheme for video flows, whose key characteristic stands in a very low storage and computational complexity. In particular, we determine the minimum amount of memory to store video traffic, based on which we can analytically determine whether incoming flows can be admitted or not subject to some predefined worst-case delay constraints. Our study relies on the framework of the deterministic network calculus, which has evolved as an attractive methodology for deriving worst-case performance guarantees in data networks.
Improving Content Delivery with PaDIS
Speakers:
Ingmar Poese
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
26 Apr. 2011
16:00–18:00
Place:
Auditorium 1
Abstract:
Today, a large fraction of Internet traffic is originating from Content Delivery Networks (CDN). To cope with the increasing demand for content CDNs deploy massively distributed infrastructures. Moreover, to minimize their cost, content delivery networks perform their own traffic optimization by assigning end-users to their servers. Such an assignment is at large unaware of the network conditions and based on inaccurate information on the location of the end-user. Thus, users are not always assigned to the CDN servers that lead to optimal end-user performance. To improve user assignment especially from a performance perspective we propose and deploy a Provider-aided Distance Information System (PaDIS). PaDIS, a novel system that allows ISPs to utilize their unique knowledge about the network conditions and user locations, helps the end-user and the ISP to improve their performance.
A Deep Dive into the LISP Cache and what ISPs Should Know About it
Speaker:
Juhoon Kim
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
19 Apr. 2011
16:00–18:00
Place:
Auditorium 1
Abstract:
Due to scalability issues that the current Internet is facing, the research community has re-discovered the Locator/ID Split paradigm. As the name suggests, this paradigm is based on the idea of separating the identity from the location of end-systems in order to increase the scalability of the Internet architecture. One of the most successful proposals, currently under discussion at the IETF, is LISP (Locator/ID Separation Protocol). A critical component of LISP, from a performance and resources consumption perspective, as well as from a security point of view, is the LISP Cache. The LISP Cache is meant to temporarily store mappings, i.e., the bindings between identifiers and locations, in order to provide routers with the knowledge of where to forward packets. This paper presents a thorough analysis of such a component, based on real packet-level traces. Furthermore, in addition to vanilla LISP, i. e., what is defined in the specifications, the implications of policies to increase the level of security of LISP are also analyzed. Our results prove that even a timeout as short as 60 seconds provides high hit ratio and that the impact of using security policies is small.
Wireless LISP: Prototyping and Evaluation
Speaker:
Marc Ferrer
Type:

Master's thesis final talk

Time:
19 Apr. 2011
10:00–11:00
Place:
Auditorium 1
Abstract:
The LISP (Locator-Identifier Separation Protocol) protocol is a network protocol that recently has been proposed as a solution for the huge growth of the Internet. The IETF has already referenced this protocol by specifying its basic principles, while it has let other domains become open subjects for further investigation. One of these subjects is the LISP Mapping System, a database system needed in the LISP implementation in order to relate identifiers with locators. In this project, a proposal for a LISP Mapping system has been implemented.
One of the subjects which still need more research in the Wireless Access Networks domain is the so-called roaming or mobility. In a WLAN formed by fixed stations (or access points) and mobile stations (or clients), the roaming process is defined by the movement of a mobile station from one access point to another due to the weakness of the signal strength received. One of the goals of this thesis is to research in the wireless mobility subject by using the LISP protocol, which can be a solution to improve mobility performance.
This project is implemented within BOWL (Berlin Open Wireless Lab), a project of the Intelligent Networks (INET) researchers group at Deutsche Telekom Laboratories. It provides an open research platform for the wireless networking community, in which several wireless networks are available for researchers to carry out investigation in the wireless networking domain. The first goal of this work is to develop a real LISP site. Hence, the innovation part of this proposal is to use this implementation carried out over wire to achieve a LISP environment in a wireless scenario. The second goal of this proposal is, once in the wireless networking domain, to carry out investigation about the mobility subject.
World of Warcraft Game Traffic Analysis based on Bro NIDS
Speaker:
Matthias Siebke
Type:

Diplom thesis introductory talk

Time:
12 Apr. 2011
16:00–17:00
Place:
Auditorium 1
Abstract:
In der Diplomarbeit Network traffic analysis of a Massive Multiplayer Online Role Playing Game (MMORPG) soll der Netzwerkverkehr des Spiels World of Warcraft (WoW), eines der populärsten Online Spiele weltweit untersucht werden.
In diesem Spiel kontrolliert der Spieler einen virtuellen Avatar in einer großen virtuellen Welt und interagiert dabei mit anderen computer- oder menschgesteuerten Avataren.
Für diese Analyse wurde einen Applicationlayer Analyzer für das NIDS Bro geschrieben, der in der Lage ist, WoW-Verbindungen zu erkennen und Informationen über die einzelnen Netzwerkpakete zu sammeln.
Des Weiteren erkennt der Analyzer in den jeweiligen Verbindungen die Bewegungspakete, d.h. Paket mit den X-, Y- und Z-Koordinaten eines Avatars, und extrahiert die darin enthaltenen Informationen.
Anhand der Daten, die durch die Analyse eines Netzwerk-Traces aus dem Jahre 2005 gesammelt werden, sollen dann in der Arbeit die Eigenschaften von WoW Netzwerkverkehr dargestellt und interpretiert werden.
Zusätzlich ist es das Ziel der Arbeit, die Bewegungsprofile von verschiedenen Avataren zu analysieren und dadurch Rückschlüsse auf soziale Bildungen innerhalb der virtuellen Welt zu ziehen.

This talk will be held in German!
Delay-based congestion control for heterogeneous environments
Speaker:
Lucasz Budzisz, FG TKN
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
5 Apr. 2011
16:00–18:00
Place:
Auditorium 1
Abstract:
In my talk I will present the details of my recent work on a novel delay-based AIMD congestion control algorithm. The main features of the proposed solution include: (1) low standing queues and delay in homogeneous environments (with delay-based flows only); (2) fair coexistence of delay- and loss-based flows in heterogeneous environments; (3) delay-based flows behave as loss-based flows when loss-based flows are present in the network; otherwise they revert to delay-based operation. I will provide an analytical proof as well as simulation results to demonstrate that these properties can be achieved without any appreciable increase in network loss rate over that which would be present in a comparable network of standard TCP flows (loss-based AIMD). Results I plan to show during my presentation will include stability and convergence results in general multiple-bottleneck networks, and a number of simulation scenarios to demonstrate the utility of the proposed scheme. In particular, I'll show that networks employing proposed algorithm have the features of networks in which RED AQM's are deployed. Furthermore, I'll show that in a wide range of situations low queueing delay is achieved irrespective of the queueing algorithm employed in the network, with only sender side modification to the basic AIMD algorithm.
Changes that the Internet brings across the concepts of publicness and privacy
Speaker:
Peter Fleischer (Google's Global Privacy Counsel)
Type:

Workshop

Time:
21 Mar. 2011
16:00–18:00
Place:
Auditorium 3
Abstract:
Discussing individual online privacy it becomes increasingly obvious that the familiar viewpoints and arguments concerning data protection are not sufficient enough to keep up with the fast-pacing development of the new media.
Do we need a new, digital understanding of privacy and publicness? Is data protection more needed than ever or rather an obsolete concept? How much transparency is acceptable and necessary for the user? What is the role of regulative institutions in that discussion? We invite you to discuss these and other issues together with Dr. Peter Fleischer, Global Privacy Counsel of Google Inc. After a short introductory statement by Mr. Fleischer we would like to start the discussion with you. We are looking forward to an interesting workshop and a stimulating debate.
Bio:
Peter works to help ensure that Google protects its users' privacy and helps to raise the bar in terms of privacy protection on the Internet. He works closely with public policy makers around the world to help update data protection concepts for the Information Age. Peter has 15 years' experience in the field of data protection, including his prior position at Microsoft. He is also a member of International Association of Privacy Professionals (IAPP). Peter was educated at Harvard College, Harvard Law School and LMU- Munich. He workes in Paris and speaks fluently German, English and French.
Prior to the event you can use the linked Moderator Tool to post your questions and comments.
Please confirm your attendance using the linked response form or via email.
NAT usage in Residential Broadband Networks
Speaker:
Fabian Schneider, UPMC
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
15 Mar. 2011
16:00–18:00
Place:
Auditorium 1
Abstract:
Many Internet customers use network address translation (NAT) when connecting to the Internet. To understand the extend of NAT usage and its implications, we explore NAT usage in residential broadband networks based on observations from more than 20,000 DSL lines. We present a unique approach for detecting the presence of NAT and for estimating the number of hosts connected behind a NAT gateway using IP TTLs and HTTP user-agent strings. Furthermore, we study when each of the multiple hosts behind a single NAT gateway is active. This enables us to detect simultaneous use. In addition, we evaluate the accuracy of NAT analysis techniques when fewer information is available.
We find that more than 90% of DSL lines use NAT gateways to connect to the Internet and that 10 % of DSL lines have multiple hosts that are active at the same time. Overall, up to 52% of lines have multiple hosts. Our findings point out that using IPs as host identifiers may introduce substantial errors and therefore should be used with caution.
NAT usage in Residential Broadband Networks (Paper @PAM '2011)
On the cost of concurrency in transactional memory
Speaker:
Srivatsan Ravi
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
8 Mar. 2011
16:00–18:00
Place:
Auditorium 1
Abstract:
The crux of software transactional memory (STM) is to combine an easy-to-use programming interface with an efficient utilization of the concurrent computing abilities provided by modern machines. But does this combination come with an inherent cost?
We evaluate the cost of concurrency by measuring the amount of expensive synchronization that must be employed in an STM implementation that ensures positive concurrency, i.e., allows for concurrent transaction processing in some executions. We consider two popular progress conditions: progressiveness and permissiveness. For permissive STMs, we evaluate the number of expensive synchronization patterns that must appear in an execution of a read-only transaction. For progressive STMs, we measure the amount of data that must be "protected" in an uncontended execution of an update transaction. For both progress conditions, our result imply that the amount of expensive synchronization needed to implement a transaction is proportional to the size of the transaction's data set.
Online Social Networks: Influence, Advertising and Privacy
Speaker:
Hamed Haddadi, RVC
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
3 Mar. 2011
17:00–18:00
Place:
Auditorium 2
Abstract:
In this talk I first present our findings about individuals' influence on Twitter as an example of an Online Social Network. I present results that indicate influence is not gained accidentally (as commonly believed by social scientists), rather it needs to be gained and maintained over time. The results also indicate that high number of followers or social links are not necessary indicative of high influence. I then present our new project, MobiAd, which is an advertising platform utilizing mobility, social network data and personal data, in a privacy-preserving manner to target the user with personalized ads.
Bio:
Hamed is a postdoctoral researcher at the Royal Veterinary College, University of London, and Department of Pharmacology, University of Cambridge. He is working on measuring cognitive, locomotor, social dynamics and behavioral functions in a transgenic sheep model of Huntington Disease. He is also broadly interested in online social networks, privacy and advertising.

A Site-specific Indoor Link Model for Realistic Wireless Network Simulations
Speaker:
Mustafa Al-Bado
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
1 Mar. 2011
16:00–18:00
Place:
Auditorium 2
Abstract:
Compared to testbeds, the efficiency and accuracy of wireless networking simulations are constantly questioned by the network community. It is widely accepted today that the current network simulators are not able to fully represent the real wireless characteristics, especially at the physical (PHY) layer. This affects the trustability of simulation-based performance evaluations. On the other hand, testbed experiments require taking a tedious and time-consuming implementation path. This path could be significantly reduced by using realistic network simulators as a first step to test novel algorithms or protocols. Therefore, we took on the challenge of representing the link characteristics of the indoor testbed of the Berlin Open Wireless Lab (BOWL) project in the ns-3 simulator. Our extensive measurements study of the link characteristics, namely received signal strength (RSS), frame detection ratio (FDR) and frame error ratio (FER), produced several guidelines for modeling our testbed with satisfying accuracy in the simulator. More importantly, the proposed empirical models take into account several crucial properties related to the radio hardware and the environment, which are shown to have a significant impact on the simulation accuracy. We validate our model against testbed results and show that, unlike the existing models in ns-3, our model shows high agreement with the measurement results for any pair of nodes in the testbed.
On Securing the Internet by Analysing and Controlling DNS Traffic and Other Activities
Speaker:
Nicolaos Chatzis, Fraunhofer FOKUS
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
16 Feb. 2011
Place:
Auditorium 1
Abstract:
The Domain Name System (DNS) is a critical infrastructure of the Internet because almost all applications that run on Internet-connected machines depend on the name resolution service it provides to work. The DNS consists of three components: the domain name space, the name servers, and the clients, formally referred to as resolvers. Due to its critical nature, the domain name space and the name servers have been for many years very attractive targets for attackers seeking to inflict widespread damage. To deal with this state of affairs, substantial attention and investment have been directed at enhancing the security of and protecting the DNS to ensure its continuous, reliable and efficient operation. This, in conjunction with a notable shift in the motivation and profile of attackers has led in recent years to a considerable change in the Internet attack landscape. Attacks have gradually become more sophisticated and focused, and financial gain has evolved into the major driving force behind them. In this new era, attackers have realised that misusing the name servers or exploiting the name resolution service comes with greater damage or economic profit than directly attacking the components of the DNS or disrupting the name resolution service. As an immediate consequence, the vast majority of Internet attacks nowadays produce an observable effect on the DNS traffic that traverses the Internet, the operation of the name servers, or in some cases on both. In the present talk, it is shown that this observation opens a new and very promising perspective for effectively detecting and mitigating a wide variety of Internet attacks.
To demonstrate the value of this perspective, the present talk is devoted to detecting and mitigating Internet worms that along with bot software are the two major Internet threats network operators and end users face. The focus is particularly on email worms, which have been, and remain, a very popular medium for attackers to achieve their ends and, therefore, the most prevalent type of Internet worms and malicious software in general. The attackers' ends include installing bot software designed to distribute unsolicited emails or launch targeted distributed denial of service attacks, stealing private information and destroying key data. In this talk, a method for detecting user machines that are compromised by email worms on the local name servers is introduced. The method uses clustering and similarity search over time series derived from the DNS query streams that user machines generate. It is demonstrated that the method overcomes the limitations of the existing methods, exhibits remarkable accuracy and negligible false alarm rate, and can be effective in the long run. In addition, a method for containing email worms is introduced. The method uses a traffic control mechanism to regulate the DNS response streams that the local name servers return to user machines and, thereby, limit the rate at which compromised user machines spread email worms further. It is shown that the method has the potential to slow down the epidemics of email worms and contribute to reducing the illegitimate email and DNS traffic compromised user machines send to the Internet with minimally, if at all, affecting their legitimate traffic.
ZOOM
Speaker:
Harald Schiöberg
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
8 Feb. 2011
17:00–18:00
Place:
Auditorium 1
Abstract:
ZOOM (Zero Overhead Online Measurement) is a tool to instrument existing distributed software to take measurements. It is designed for extremely high measurement event rates, as it is originally designed to take per-packet measurements of the internal state of the Click Modular Router framework, yet it's not at all limited to Click. I will present the design, the state of the current implementation and early performance results, and hope for a good discussion about possible applications of this tool to other high-volume measurement tasks.
The Windfall of Friendship – Altruism in Game Theory
Speaker:
Yvonne Anne Pignolet, IBM Research / ABB Research
Type:

Talk

Time:
4 Feb. 2011
10:00–11:00
Place:
Auditorium 2
Abstract:
In this talk we consider a virus inoculation game on social networks. A framework to capture the windfall of friendship is presented, i.e., it measures how much players bene fit if they care about the welfare of their direct neighbors in the social network graph compared to purely sel fish environments. Intriguingly, even though the windfall of friendship can never be negative, the social welfare does not always increase monotonically with the extent to which players care for each other. While these phenomena are known on an anecdotal level, we can quantify these effects analytically with the help of this framework
Toward complex network models suited to real world networks
Speaker:
Gilles Trédan
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
25 Jan. 2011
16:00–18:00
Place:
Auditorium
Characterizing tradeoffs in split architecture
Speaker:
Dan Levin
Type:

Project group meeting (Projektgruppentreffen, PGT)

Time:
18 Jan. 2011
16:00–18:00
Place:
Auditorium
Abstract:
Split architecture networks, e.g., Openflow, introduce tradeoff problems concerning how reactively or proactively the control plane interfaces with the data plane. Purely reactive networks offer high degrees of flexibility in making forwarding decisions, but introduce problems for network scalability (in terms of network size, amount of control traffic overhead, network latency performance, possibly other areas as well). As networks become more proactive, the flexibility and degree of control over forwarding decisions decrease, but scalability and performance should increase.
The characteristics of a given network and the traffic demands placed on it play a significant role in determining how reactively or proactively the controller-dataplane interaction is managed. Thus, we want to first consider, what circumstances govern how reactively a network be operated. Secondly, as we change the measure of network reactiveness, how do the characteristics of that network change, e.g., how does the control traffic overhead grow or shrink, how much visibility into the data traffic do we gain or lose from the control traffic. As we begin to characterize these reactivity tradeoffs, we believe more general trends will emerge which may be applicable to other networking areas. we aim to present our early plans and solicit feedback

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions