Technische Informatik - Network Organization and Protocols (2,3,N)
This seminar is a block seminar, i.e., at the end of the semester, there will be two days when all talks are given in a block. During the semester, there will be deadlines for status reports, but no meetings of the complete seminar group.
Students participating will give a talk and write a report (7 pages as a general rule but each advisor may specify his own requirements). Besides that, they will review one or two reports written by other students and prepare questions to be asked after the corresponding talk. This is to stimulate discussions after the talks and make the seminar more worthwhile.
|Dozent/in||Prof. Dr. Mesut Günes|
|Institution||Institute of Computer Science|
Freie Universität Berlin
|Raum||Takustraße 9 K 40, SR 51|
Friday, 09.07.2010, 14:00 - 18:00, K 40, Takustr. 9
Part I: Meet (12.04. - 16.04.2010)
- [15.04] Organizational Meeting, 15:30, SR 051, Takustr. 9
- [16.04] Send a list with your five preferred topics in the order of declining preference to seminar-agtech[at]inf.fu-berlin.de
- [18.04] You will receive your topic via email
- Please read the literature linked below!
Part II: Preparation (19.04. - 23.04.2010)
- Meet your advisor to discuss content and structure of your report
- [23.04] Participate in "Introduction to scientific writing", Room K40, 13:00
Part III: Write Seminar Article (19.04. - 18.06.2010)
- Read, discuss, write, discuss, re-write your seminar article in tight communication with your advisor
- [18.06] Submit your seminar article per email to seminar-agtech[at]inf.fu-berlin.de
Part IV: Review (14.06. - 02.07.2010)
- [21.06] You will get 3 seminar articles for review
- Read the seminar article and write a review for each
- [25.06] Submit your reviews per email
- Incorporate the comments from the reviews into your seminar article
- [02.07] Submit your final seminar article per email to seminar-agtech[at]inf.fu-berlin.de
Part V: Presentation (05.07. - 09.07.2010)
- [07.07.] Presentations (first part), 16:00 - 20:00, SR 053, Takustr. 9
- [09.07.] Presentations (second part), 14:00 - 18:00, K 40, Takustr. 9
- Taking part in the final presentations is mandatory!
Attention: Students have to meet all deadlines listed in the schedule. Otherwise, she/he will lose the right to take part in the final presentation.Deadlines are specified in brackets.
Schedule of final presentations
|Mi. 16:10||BGP Prefix Hijacking||Christopher Flach||Matthias Wählisch|
|Mi. 16:45||Mobile Sandbox||José Manuel Barroso Galindo||Matthias Wählisch|
|Mi. 17:20||6LoWPAN||Oliver Gesch||Oliver Hahm|
|Mi. 18:00||Gossip Routing||Michael Busch||Bastian Blywis|
|Mi. 18:35||Current Consumption in WSNs||Olaf Loga||Norman Dziengel|
|Mi. 19:10||Sensors and multiple Sensors in WSNs||Lutz Vöhringer||Norman Dziengel|
|Fr. 14:00||Mobility models for human movement patterns||Karl Held||Felix Juraschek|
|Fr. 14:35||Analyzing computer networks using network calculus||Yosif Harizanov||Katinka Wolter|
|Fr. 15:10||Gossip-Based Data Distribution||Mathias Hecht||Katinka Wolter|
|Fr. 15:55||P2P Data Storage||Daniel Steig||Kaspar Schleiser|
|Fr. 16:30||Adaptive Hybride Funkortung||Miriam Ney||Zakaria Kasmi|
|Fr. 17:05||Dead Reckoning||Andreas Franz||Zakaria Kasmi|
Advisor: Matthias Waehlisch
Digital Immune System on Mobile Devices
Without doubt, mobile, wireless-based end devices represent a significant part
of our current networks, both with respect to the number of deployed equipment,
and their economic impact. Limited hardware resources and the per se
unprotected transmission media air turn such devices into attractive targets
for attacks. Mobile end devices, thus, need a lightweight, environment-adaptive
protection mechanism that exploits the heterogeneous, technologies available
on-board. The aim of this work is to exploit current solutions for a digital
immune system on mobile devices. The literature survey should classify the
approaches around. This includes the layer on which the system is protected and
if the protection is cooperative.
Assigned to: Hernando Saen Sanchos
A sandbox is a security approach to run programs in a separated environment on
the local system. Programs executed in a sandbox should be isolated and not
conflict with other processes. This mechanism, thus, allows for a safe test or
run of untrusted software. Today, sandboxes are well-known and established for
fixed environments. The aim of this work is to exploit research on mobile
sandboxes. Are there any specifics that applies for mobile devices (e.g.,
mobile phones)? Are there any approaches around, which work towards distributed
Assigned to: José Manuel Barroso Galindo
BGP Prefix Hijacking
In the current Internet, inter-domain routing is based on BGP. A BGP router
advertises IP prefixes that it is responsible for. However, the model how ISPs
share BGP routing information implicates inherent security problems. On the one
hand no one prevents a BGP speaker to announce an arbitrary prefix, on the
other hand there is no straightforward way to shield BGP routers from invalid
updates. In February 2008, one of the most prominent cases occurred when
Pakistan Telecom started an unauthorised announcement of an IP prefix that
belongs to YouTube. This incorrect update has been globally propagated.
Although Pakistan aimed to block the YouTube website only locally, this
accident interrupted YouTube traffic around the world. The aim of this work is
to look on current studies of prefix hijacking and to discuss tools and
solutions to overcome this problem.
Assigned to: Christopher Flach
Advisor: Marco Ziegert
Compression techniques for analog data in WSNs
Capturing analog data on a sensor node, like temperature, light intensity,
distances or even sound may be a normal use-case in wireless sensor networks.
However, this can be a challenge if multiple sensor nodes compete at the same
time to send their raw data. In these cases, it may be useful to make
sacrifices on data quality and use a varying compression level, dependent on
current traffic from source to sink to raise the transmission probability.
Present and value (non-)lossy data compression techniques that focus on analog
data and also qualify for the restricted resources on low-power sensor nodes.
Give a hint on providing a higher compression ratio on already compressed raw
data if estimated traffic differs while data is transferred to the sink.
Detecting obstacles in a WSN area out of long term monitoring
Maps created out of RSSI distance estimations only give a small hint on a real
scenario like an office building. Obstacles like walls distort these
measurements. Assuming a normal WSN whose sensor nodes also have temperature
and humidity sensors, these data can help identifying possible distortions and
improving the data quality and finally the map. Imagine two rooms with at least
1 sensor node, a person opens a window in one room, the temperature and
humidity decrease whereas room no.2 is not affected. A wall between both
sensors nodes might be quite likely and the previous distance estimation
between them might be to high or completely wrong. Present and value past and
current studies and create an own model, based on only a limited number of
sensors, preferably compatible to those on WISEBED sensor nodes.
Advisor: Norman Dziengel
Braided Multipath Routing in Wireless Sensor Networks
Wireless Sensor Networks are still lacking on the perfect routing that helps to
extend the lifetime of the network itself. The problem is simple: A good
routing algorithm will find a proper route to the sink. In the case of a stable
route, this route will be used very often. Hence, the energy of all nodes
involved in this route will deplete there energy much earlier than other nodes.
As the WSNs aim to extend their lifetime, it is not recommended to reuse a
working route for a long period. A solution could be the usage of alternating
routes that do not use nodes that are elements currently active routes. Is it a
good idea to use completely distinct routes in an alternating way? Maybe it is
better to establish routes differing in only one node? How is the node's
memory affected and how is data between the nodes affected, as it is necessary
to store alternate routes. Compare given approaches and find a conclusion.
Give a clear view into the current state of the art.
Assigned to: Olaf Loga
Current Consumption in Wireless Sensor Networks
Wireless Sensor Networks are still lacking on the perfect routing that helps to
extend the lifetime of the network itself. Ideally, we know the state of energy
within all nodes. A solution could be to measure the energy within the nodes
with appropriate heuristics. But how precise is the community working?
Sending/Receiving a packet is expensive as the transceiver requires most energy
of all components. Find the second worst component in a WSN next to the
transceiver concerning current consumption. Which methods are known to
precisely measure the current consumption? How is the measuring itself
influencing the current consumption? Are there other approaches than measuring
and approximation?Find real world scenarios approaches and compare their work.
Give a clear view into the current state of the art.
Sensors and Multiple Sensors in Wireless Sensor Networks
Measuring data in a Wireless Sensor Network is very important as most
applications need to know some parameters about the environment to get an
impact out of a WSN. Hence the quality of a sensor is important for proper
measurements and the subsequent evaluation. What does it mean to calibrate a
sensor like a temperature, humidity, IR, US or acceleration sensor? Commonly, a
sensor is calibrated after buying it, but its data-sheet gives information
about its error types. Is it possible to raise the quality of a measurement by
using more than one sensor of exactly the same type, or is it more advisable
to use different kind of sensors, e.g. of different merchants. Find approaches
that compare sensors or that try to raise the resulting measurement by using
multiple sensors. Evaluate the approaches and give a clear view into the
current state of the art.
Assigned to: Joschy Vöhringer
Advisor: Waqaas Munawar
Cost of Security in WSNs
Security, like friction, is a necessary evil. WSNs are mostly foreseen to be
used in: disaster relief situations, body area networks, assisted living,
building automation, environmental sensing and reconnaissance. All of these
require different levels of security integration. Hence the cost of the
security, the user is willing to pay, differs as per the situation the network
is deployed in. This work shall report on the security requirements of a WSN
and how much does it cost to fulfill these in terms of processing time, energy
toll, communication overhead and programmer's effort.
Ideal Code Dissemination Protocol
Post deployment software updates are frequent in Wireless Sensor Networks
(WSNs). These can result from a variety of factors including bug fixes,
interest shift or post analysis experiment redesign. Once the nodes have been
deployed, collecting them again to apply a software update and then redeploying
them is not only a hassle but sometime hazardous as well. Therefore, remote
retasking is one of the fundamental requirements of a practical WSN. On the
other hand, communication is the most draining activity for sensor nodes hence
is a prime target for optimization. In this task the existing protocols for
code dissemination shall be analyzed and the requirements of an ideal protocol
shall be formulated.
Advisor: Zakaria Kasmi
Die Navigation spielt in vielen Bereichen eine immer größer werdende Rolle.
Diese Navigationssysteme werden als unterstützende Systeme im Automobilsektor,
autonome fahrende Geländewagen oder agierenden Robotern im Bereich der
Fördertechnik oder in der Logistik.
Ein Dead Reckoning System ermöglicht die Lokalisierung und die Navigation von
Objekten oder von Fußgängern innerhalb von Gebäude, neben den drahtlosen
basierten Lokalisierunksysteme (WLAN, UWB, GNSS, etc).
Es basiert auf der Laufrichtungsmessung (Kurs) und des durch ein Objekt
Ziel dieser Arbeit ist die für das „Dead Reckoning“ verwendeten Verfahren sowie
die auf dem Markt vorhandenen Systeme vorzustellen.
Assigned to: Andreas Franz
Adaptive Hybride Funkortung
Die Lokalisierung von Objekten und Personen im „Indoor“-Bereich spielt bei vielen Anwendungen eine große Rolle. Die wichtigsten Anwendungsgebiete eines Indoor-Positionierungssystems sind:
* Auffinden von Objekten: z.B. eine benötigte Ressource, den nächsten Drucker, oder einen Mitarbeite.
* Umgebungssteuerung: Steuerung der Klimaanlage und der Beleuchtung in Abhängigkeit vom Aufenthaltsort der in einem Gebäude befindlichen Personen.
* „Augmented Reality“: Anreichung der realen Welt mit virtuellen Objekten, die beispielsweise über eine Cyber-Brille optisch dargestellt werden. Dafür werden eine genaue Positionierung sowie eine Orientierung des Blickbereichs benötigt.
* Drahtlose Sensornetze: Zusätzlich zu dem Erwerb, Weiterleitung, Verarbeitung und Bereitstellung von Informationen aus der physischen Welt, ist eine Feststellung der Position eines Sensorknotens in einem räumlich verteilten Sensorknotensystem sehr nützlich.
Die meisten vorhandenen Systeme sind Insellösungen, die nur eine einzige Funktechnologie verwenden. So können Signalschwankungen und Abdeckungslücken sehr schwer kompensiert werden. In diesem Fall ist die Positionsgenauigkeit von der Raumtopologie und den Technologiegrenzen des verwendeten Verfahrens sehr abhängig.
Bei einem hybriden Funksystem verwendet man gleichzeitig unterschiedliche Verfahren zur Ortsbestimmung, mehrere Strategien mit unterschiedlichen Genauigkeiten und eine adaptive Einbindung verschiedenerer Positionierungstechnologien.
Ziel dieser Arbeit ist es, nach Verfahren zu suchen, die die oben genannten Kriterien erfüllen können sowie die bis jetzt vorhandenen hybride Funksysteme.
Assigned to: Miriam Ney
Advisor: Bastian Blywis
Evolutionary Routing Algorithms
Evolutionary algorithms are inspired by the biological evolution. They use
reproduction, mutation, selection, and recombination of a population of
entities to solve problems that are considered hard. The individuals are
evaluated by a given fitness function which determines their survival. After
several generations, one individual with the highest fitness will represent a
solution to a particular problem. This solution might be sub-optimal but a
good and satisfying approximation.
This approach can also be applied to routing in networks as other researchers
have already proposed. In this seminar topic you shall introduce the topic of
evolutionary algorithms, their advantages and disadvantages, and show how this
approach can be applied to routing in computer networks. You have to discuss
if the presented algorithms are centralized or fully distributed and if they
can be applied in all scenarios.
Flooding is a very simple mechanism, that is applied in routing protocols at
least in the initial stage. To minimize overhead while still retaining full
reachability, gossiping, also called gossip routing, can be used. In contrast
to flooding, where packets are always relayed as long as the time to live (TTL)
value does not reach zero, the forwarding is probabilistic. A node forwards a
packet with a given probability p or drops it with 1-p. In the last decade
several gossip routing variants have been proposed. In this seminar topic you
have to introduce several variants and discuss how they perform considering the
three most important metrics: reachability, redundancy, and overhead. You have
to provide application scenarios resp. example network topologies, where
particular gossip routing variants will perform badly.
Assigned to: Michael Busch
Advisor: Oliver Hahm
The integration of the so called Internet Of Things into worldwide network
structures is one the big challenges for the developement of the Future
Internet. Tiny low-power devices such as wireless sensor nodes and embedded
systems that are used in Personal Area Networks (PANs) have to been connected
to well-established network topologies.
Using IPv6 over a 802.15.4 based MAC layer is one of the possible approaches to
solve this problem. For this purpose IETF has created a working group called
6lowpan. You should give a detailed overview about the topics discussed by this
working group like bootstrapping, header compression and routing.
Assigned to: Oliver Gesch
Routing protocols supporting mobility
To meet the specific requirements of a mobile ad hoc network (MANET) a couple
of routing algorithms have been developed during the last decades. On of the
challenges of these protocols in particular is obviously the mobility of the
nodes. As a goal of this assignment you should create a survey over the
different approaches to deal with this topic. Which routing protocols exist and
what is the current state of their implementations? Which strategies are used
to handle mobile node in such a network?
Assigned to: Jun Bao
Advisor: Felix Juraschek
Mobility models for human movement patterns
Network simulations for mobile wireless networks usually rely on mobility
models to specify the movement of the mobile entities. These models describe
movement patterns that are characteristic for the particular mobile object.
The focus of this topic is on models for human movement patterns. Simple
mobility models, such as the Random Waypoint model, are easy to calculate and
therefore scale very well. However, the derived mobility traces do not
resemble actual human movement. More complex models have been developed that
model human and group movement patterns in urban environments. Candidates of
these models are variations of the Levy Walk model, models based on real
mobility traces, and models that respect geographic restrictions.
The goal of this thesis is to acquire an overview of mobility models for human movement
patterns and derive a discussion about their features and limitations.
Assigned to: Karl Held
Interference metrics for wireless networks
Interference in wireless networks occurs, when two near-by network stations
transmit simultaneously on the same frequency and thus may affect the receiving
process. The research area of channel assignment attempts to decrease the
network-wide interference by using non-interfering channels for potentially
interfering links. An interesting task is how to estimate the interference and
its effects on the network performance, since it is usually of a dynamic nature
and the intensity may change over time. Interference metrics have been
developed for this case. These metrics range from simple ones which specify an
interference range, usually the m-hop neighborhood of a network station, and
all links in this range potentially interfere with each other. More complex
models estimate the interference of two links with measurements, which is a
time and resource-intensive procedure but promises more accurate results.
However, the gathered results are usually only valid for one particular network
deployment and can thus not be transferred to other networks.
The focus of this topic is to acquire an overview of the state of the art of
developed interference metrics and derive a discussion of their applicability
for particular scenarios.
Assigned to: Robin Pradel
Advisor: Katinka Wolter
Analysing computer networks using network calculus
Network calculus is a mathematical description that allows to analyse networks
with respect to throughput, bottlenecks and performance bounds. Traffic flows
in a network can be described as they travel through a network. The
presentation and report should introduce and explain the method, demonstrate it
in an example and point out strengths and weaknesses.
Assigned to: Yosif Harziano
GMAC: A Game-theoretic MAC Protocol for Mobile Ad Hoc Networks
gMAC is a protocol developed by the Dutch company CHESS for decentralised
medium access in wireless sensor networks. These networks are used e.g. in a
Dutch flower market where thousands of trolleys are autonomously routed to
transport flowers. The protocol is decentralised and energy-aware. The
presentation and report should introduce and explain the protocol as well as
consider evaluation of the protocol.
Advisor: Kaspar Schleiser
P2P Data Storage
Modern P2P Data Storage Systems like Pastry, Chord or Freenet make use of
sophisticated algorithms to store and retrieve data in massively scalable P2P
networks. Most are based on Distributed Hash Tables (DHT). Redundancy makes it
possible for nodes to leave the network without data loss. Most systems offer
some kind of encryption and some even allow participants to remain anonymous.
he focus on this topic is to acquire an overview of currentl state of the art
protocols regarding features, performance, scalability and resource usage.
Assigned to: Daniel Steig