Master/Bachelor Thesis - Proposals
Web-based real-time communication (16hp)
Contact: Simin Nadjm-TehraniRobot-assisted search and rescue operations can be improved when adding extra sensing capabilities.
Availability of embedded cameras can enhance robot location awareness but can also enable guided search with remote user interaction when sending visual information over a communication network.
Real-time video streaming to a remote server, however, is a challenging aspect in environments where connectivity can be disturbed or the bandwidth overloaded.
Quality of Service (QoS) adaptation is required in order to adjust the video quality to the level of available resources (bandwidth, CPU, etc.)
The aim of this project is to develop an application that allows a robot to
transmit video and sound from the area that it is visiting to a remote server, displaying it on a screen at the server end. The application should be developed by using the WebRTC suite, a recent API framework for real-time communication in web applications, and should be designed with QoS adaptation capabilities.
Survey/demo of vulnerabilities of mobile platforms due to NFC interfaces (16hp)
Contact: Simin Nadjm-TehraniThe handheld devices and smartphones are being used in novel settings including some that are dependent on the use of short range radio for particular applications, e.g. payments over dedicated terminals using near field communication (NFC). The goal of this Bachelor thesis project (16 hp) is to study the security vulnerabilities and threats that the NFC communication creates seen from the perspective of a mobile user. The problem can be studied in a theoretical fashion by studying available published material on such threats, as well as creating a testbed in which local attacks on a given (test) device are created and demonstrated by the student.
From location-tagged social interactions to mobility models (30hp)
Contact: Simin Nadjm-TehraniThe mobility of individuals over extended periods of time can increasingly be attached to their interactions over social networks. There are several applications (such as Twitter) that make tracking of the individual’s locality possible through tagging the social interactions with location data. This master thesis project consists of two components: 1) a study of the state-of-practice in the social applications domain and identifying the extent of availability of the location data over time for a subset of interaction traces that are openly obtainable on the Internet. 2) using the traces collected to create typical (generic) movement patterns that one can use to test applications that are dependent on mobility models. For example dissemination of some attractive news, or a warning, on a peer-to-peer basis over a geographic area, and computing the latency for the dissemination based on the generic movement patterns (so called mobility models).
Vehicular information delay
Contact: Mikael AsplundSignificant research efforts are currently being devoted by both
industry and academia on getting groups of cars to collaborate using
wireless communication. Such collaborative algorithms have the
potential to save thousands of lives every year as well as to increase
the efficiency of traffic and reduce the environmental impact of the
transportation sector. However, poor reliability of the communication
channel and difficulty in guaranteeing timing properties remains a
major hindrance. This project aims to contribute to the creation of
more reliable information dissemination mechanisms by providing a
basic delay prediction mechanism which is able to forecast the
communication delay experienced by a connected vehicle in an urban
scenario. The student will use a simulation platform to measure timing
characteristics of wireless communication for a given vehicular
mobility model. This will allow creating a basic model of how the
information delay varies with factors such as distance and node
Large scale simulations
Contact: Mikael AsplundVehicular networks are emerging as a hot research topic with prototype
deployments already being tested in the US. One major challenge for
the research community is to simulate large scale systems with tens of
thousands of cars. Today, most simulation-based studies are restricted
to a few hundred cars or less. The purpose of this project is to
investigate the main bottlenecks in simulating large scale systems for
an open source network simulation platform. Once the main performance
issues have been found, the student will implement and evaluate
modifications to the simulation platform to radically increase the
number of cars that can be handled by the simulator.
Energy optimization for embedded telecom applications
Contact: Simin Nadjm-TehraniToday's computing platforms (e.g. Linux-based laptops) have a built-in mechanism for adapting the energy consumption during computations, by switching to low power modes with lower frequency or voltage (so called frequency/voltage scaling). These mechanisms have been created to run with arbitrary applications and are thereby only partially adapting to what is going on during a computation. In an embedded system, e.g. a telecom base station controller, the program running on a platform is a dedicated program. Its code is well-known and one can parameterize and optimize its "duty cycle" depending on load, where in the control flow the program currently runs, and its expected near time computational needs. The goal of this project is to study how a new layer of energy optimization can be placed over the generic Linux mechanism and (preferably) without changing a telecom application code. An example application can e.g. be in the language Erlang. The candidate student (either one Master level student, or two Bachelor level students working together) needs to have taken a concurrent programming or operating systems course, and has programming capabilities needed to create a layer that interacts between the Erlang run-time environment and Linux frequency/voltage scaling mechanisms. This work will be done in cooperation with Ericsson Radio systems at Linköping.
A secure BATMAN implementation for the Android platform
Contact: Simin Nadjm-Tehrani
Traditional internet based networks are composed of many networks with a number of core nodes. These nodes, which are owned by several cooperating entities, route the data according to pre-established agreements. As opposed to this, in mesh networks, there are many nodes than do not only transmit their own data, but they also forward the data of other nodes. The information routing in mesh networks is not based on agreements, but on dynamic routing or dissemination protocols. These protocol can be classified under "ad-hoc networking".
One example protocol is BATMAN, a totally decentralised routing protocol that dynamically creates the routes to forward the data. It was originally intended for communication in rural areas where commodity mobile handsets can set up a network with no pre-existing infrastructure.
The decentralisation and the adaptability to network topology changes are the main advantages of BATMAN. Nevertheless, the information exchanged among the nodes to keep the routing tables is not protected and makes it vulnerable to attacks. The purpose of this thesis project is to 1) implement the BATMAN protocol on an Android platform (or find open source code that already does it). For this existing (C++) code on a simulation platform in our lab can be used as a basis. 2) Reduce the security threats by enhancing BATMAN with authentication based on X.509 certificates on the same platform. For this part existing code from another Master student project can be integrated.
Fault and recovery propagation in automotive embedded controllers
Contact: Simin Nadjm-Tehrani
Systems that are operated in adverse conditions and use non-expensive components are also likely to suffer the impacts from transient and permanent faults, resulting in multiple (degraded) modes of operation for each of the components. The existing version of EMS in Scania trucks includes on average 3 modes for each of the involved components. In order to choose the right mode of operation for control modules, the components and derived signals thereof have to be continuously diagnosed, the results of diagnostic tests recorded on-line, and later revised when the fault is corrected; either by the service mechanic, an action by the driver, or by a physical condition that reverses/eliminates a transient.
This Masters thesis addresses the architectural decisions and interaction models among controller components such that desirable fault propagation and recovery sequences can be verified and enforced. The work includes the study of the tools associated with AADL (Architecture Analysis and Design Language) for modelling and analysis of such propagation flows in automotive applications.
This project is performed in cooperation with Scania (truck manufacturers) in Södertälje.