AAU logo
One or more PhD Stipends in Electronic Systems

One or more PhD Stipends in Electronic Systems

Application deadline is June 15, 2020

Read more 


Project Proposals

All candidates must select two projects (1st and 2nd priority) from the list of the 24 potential PhD topics shown below. The choice of projects must be clearly indicated and uploaded together with the appliccation and a one-page description of the research as envisioned by the candidate.


1. BED: Botnet Economic Disruption via dark web marketplace infiltration

Every year, cyber-attacks result in costs of hundreds of billions of dollars worldwide. One of the core facilitators of attacks are botnets, which are networks of infected computing devices. This PhD project will investigate a novel interdisciplinary approach for defending against botnets. BED will utilize research from the fields of network security, business economics and law, along with specialized research in botnet monitoring.

The goal of BED is to

identify and infiltrate botnet marketplaces with an emphasis on the dark web,
model the business economy of botnets,
Analyse the influence of botnet-specific parameters to the economy model and iv) propose a botnet economy disruption plan.

The PhD candidate will join the growing AAU cyber-security network in Copenhagen and work together with an interdisciplinary group of experts. The candidate will be able to collaborate with leading security experts worldwide and present their findings in top conferences all over the world.

Supervisors: Jens Myrup Pedersen, Emmanouil Vasilomanolakis and Morten Falch

Workplace: Campus Copenhagen


2. Cloud Robotics: Distributed allocation of resources for wirelessly connected mobile robots

Cloud robotics is a new paradigm that may finally get the robots to work together. Easy access to computing power in the cloud opens the possibility of moving intelligence from individual machines into the cloud. This enables new cooperative applications, as interconnected robots can complement each other’s sensing and act in concert to reach goals that are too complex for any single robot. Such a transition puts high communication demands that can be addressed by the emerging 5G wireless technology. This spawns many challenging questions, such as:

How is the software architecture structured?
How are the robots capabilities dependent on the state of communication system in terms of data loss, latency or similar?
How should the robot and communication system adapt to the current state of the environment?
How can multiple robots benefit from distributing the intelligence into the cloud?

Supervisors: Petar Popovski, Karl Damkjær and Federico Chiariotti

Workplace: Campus Aalborg


3. Robust Control Coordination of Distribution Grid Connected Systems using intermittent ICT connectivity

This PhD focuses on the interlink between communication and control of energy distribution grids in a set of use cases including voltage and reactive power control, grid efficiency optimization, and control coordination across different control targets. These use cases and related functionality rely on ICT infrastructure that ensures data access and interfacing to various subsystems (e.g. inverters, storage), which each have different access characteristics and properties. Control coordination applications for these use cases will need to be robust towards varying communication properties, data quality, faults and other limitations, thus the applications will need to be aware of and interact with the underlying ICT efficiently to avoid spread of failures in subsystems. The goal of the PhD will be to ensure that control and coordination functionality from several applications can be executed despite an unreliable ICT infrastructure and that it co-exists with other control applications.

Supervisors: Rasmus Olsen and Jan D. Bendtsen

Workplace: Campus Aalborg


4. Detection of Cyber Attacks

Malfunctions and cyber-attacks on the components have more often than not devastating effects on critical infrastructure. Therefore, any anomaly has to be rapidly detected before it evolves into an accident. For a cyber attack, this translates into detecting the attack in the earliest stages of infection - before the malicious activity is carried out. 

This PhD project aims to develop theory and tangible algorithms for the detection of malfunctions and cyber-attacks. The idea is to use the information about network activity and, most importantly, to employ a physical model that predicts the expected evolution. If a fault occurs, the discrepancy between the prediction and real situation is observed. This deviation is used for the early detection and isolation of the point of attack.

We anticipate computing the probability of the attack and use classification techniques known in control theory and machine learning to determine the software-code under attack.

Supervisors: Rafael Wisniewski and Jens Myrup Pedersen

Workplace: Campus Aalborg


5. Deep Learning Based Communication for Power-Efficient Satellite Systems

Satellite communication has drawn great attention in recent years due to the rapid development of low Earth orbit (LEO) satellite constellations to provide global satellite Internet access. LEO satellites have the features of low altitude of 500-1500 km and short transmission windows (approximately 10 minutes per pass) with the ground stations, which leads to the stringent need of high data rate communication schemes. This project aims at deriving new communication techniques based on deep learning to improve the overall performance of satellite communication systems. The special focus includes the reduction of peak-to-average power ratio (PAPR) and nonlinear distortion of multi-subcarrier signals to gain high power efficiency in the emerging large-scale low Earth orbit (LEO) satellite constellations. Autoencoder architectures using deep neural networks (DNNs) will be adopted as the main solution to achieve the goal.

Supervisors: Ming Shen and Elisabeth De Carvalho

Workplace: Campus Aalborg


6. Deep Multi-Parameter Fingerprinting for Enhanced Device Authentication in IoT Healthcare

With the increasing popularity of the Internet of Things (IoT) for healthcare, the risk of healthcare data breach is growing, leading to the need of stricter authentication of devices before allowing them to access the network. The goal of this project is to develop a multi-parameter fingerprinting system for device identification and authentication enhancement in networks used for IoT healthcare. Compared with conventional data-driven approaches, using, e.g., IP or MAC addresses, fingerprinting of user agents, and client IDs, this project proposes the use of hardware fingerprinting at the base station to screen and identify all devices before allowing them to connect to the network. This is achieved by using deep learning techniques (e.g., convolutional neural networks) to identify user devices (e.g., phone, tablet or laptop) based on their hardware-dependent radio frequency (RF) “fingerprint” (e.g., waveforms, frequency or phase offsets), which is difficult to imitate.

Supervisors: Ming Shen, Henning Olesen and Reza Tadayoni

Workplace: Campus Aalborg


7. Deep Wireless and Visual Perception for Autonomous Navigation of Robots

In the past decades, we have seen a rapid evolution of robotic technologies and large-scale deployments of robots in different applications (e.g., car factories) for tasks involving high levels of repetition, intense labor, or high precision. As the tasks for next-generation robots will become more challenging than simple repetitive tasks, the need of higher autonomous level is also urgent. Autonomous navigation is the most fundamental component of robotic autonomy and hence has attracted significant attention. The goal of this project is to investigate and develop new navigation techniques for next-generation autonomous robots. Wireless radio frequency sensing signals (mm-wave radar and UWB) and vision perception information (Camera, Laser) will be concurrently utilized by deep learning techniques for global localization (e.g., smart routing) and local environment positioning (e.g., obstacle avoidance), with a special focus on improving the localization precision and robustness.

Supervisors: Ming Shen and Rafal Wisniewski

Workplace: Campus Aalborg


8. Signal Processing Enabled Intelligent Antenna Array Testing

Advanced antenna technologies, including large-scale antenna configuration and utilization of millimetre-wave bands, are seen as key technologies to enable future communication systems.  Consequently, large-scale integrated antenna systems are now conceived, where massive antenna elements equipped with integrated radio frequency circuits are combined in one physical unit. Before such antennas can be deployed, they should be fully tested to see if specifications are met. However, such advanced antenna technologies make traditional antenna testing methods impractical, expensive, time-consuming, and labour-intensive.  In this project, we aim to develop and experimentally validate automated and fast signal processing enabled solutions for antenna array testing, e.g. radiation characteristics acquisition, array calibration and faulty element(s) detection.  The project will be based on a steady interplay between mathematical modelling of antenna measurement system and design of algorithms.

Supervisors: Wei Fan and Troels Pedersen

Workplace: Campus Aalborg


9. Machine Learning Assisted Meta Surface Design and Optimization for High-Performance Millimetre-Wave Transmit Array/Reflect Array Antennas

In this proposal, designs and optimizations of meta-surfaces enabled by machine learning for high-performance millimetre-wave transmitarray/reflectarray antennas are proposed and described for the upcoming 5G/6G millimetre-wave applications. The performance of such antennas is mainly dependent on the phase distributions of meta-surfaces. From the currently-reported work, it’s time-consuming and non-robust to determine the phase distributions of a meta-surface to achieve the desired performance. Machine learning, as a powerful optimization technique, can be efficiently adopted to optimize the phase distributions of meta-surfaces for versatile properties, such as wideband, high aperture efficiency, low profile, low sidelobe, et al. The proposal is aimed to the current technical difficulties in industry and academia, designing and optimizing three kinds of meta-surfaces with machine learning technology to achieve an antenna with wideband, low sidelobes, and multiple beams, respectively.

Supervisors: Shuai Zhang and Zheng-Hua Tan

Workplace: Campus Aalborg


10. Reliable long-range operations of unmanned aircraft by means of jointly optimized guidance, navigation, and communications (LongRangeDroneComm)

Long range drone flight (20+ km), requires uninterrupted radio contact to ensure high safety operation. Radio interruptions are often caused by the aircraft masking the signal path between drone and ground station. The goal of the project is to devise a holistic methodology for optimizing communication link quality during a flight mission through the joint optimization of a) drone movements and orientation, b) positioning of the antenna, and c) protocol/radio interface adaption, all while being subject to the constraints of the flight mission. Specific tasks are:

implementing a joint communication link quality and aircraft attitude simulator,
analysing the effect of antenna location vs. aircraft attitude,
flight mission planning subject to attitude restrictions,
investigate the potential of using multiple communication interfaces, and
participation in flight tests in collaboration with AAU Drone Research Lab to experimentally validate the proposed feature enhancements

Supervisors: Jimmy J. Nielsen, Anders la Cour-Harbo and Ming Shen

Workplace: Campus Aalborg


11. Assessment of health effects due to 5G mobile phone radiation using oto-acoustic emissions

In the age of mobile communications, there is a recurrent discussion about the existence of possible harmful health effects from mobile phone radiation. Despite this interest, there is still a lack of knowledge and reliable methods to assesses possible effects on the human brain. Oto-acoustic emissions (OAE) reflect the state active processes of the inner ear controlled by efferent fibres of the auditory pathways. OAE have proven useful in the assessment of temporary changes in hearing due to sound exposure and oto-toxic drugs. So, if 5G mobile phone radiation produces changes in the normal neural activity in the brain, then these changes could be observed in OAE measures. The goal of the PhD project is to investigate viability of using OAEs to determine possible health effects of 5G mobile phone radiation. The research will include: identifying relevant OAE measurement paradigms, laboratory and field investigations of relevant exposure conditions, and experiments with human subjects.

Supervisors: Rodrigo Ordoñez and Ondrej Franek

Workplace: Campus Aalborg


12. Robot Personality Framework Canvas

Human-Robot Interaction (HRI) is the study of interactions between humans and robots. It is a multidisciplinary discipline consisting of research from engineering, computer science, psychology, etc. The goal of this project is to develop a Robot Personality Framework Canvas, which can used by different disciplines to scope the development of social robots for HRI. The project will employ a user-centric iterative research strategy. Real world studies will be conducted using robot research platforms and prototyping techniques.

The main research questions are:

Which modalities are central for social robots in a specific set of uses cases and how are these correlated?
Which methods can be used for scoping the physical design of social robots and for selecting their behavioural movements?
How to translate findings from user experiments to an engineering context and vice versa?
How to scope and convey bridge building between explorative studies and system developers?

Supervisors: Rodrigo Ordoñez and Karl Damkjær

Workplace: Campus Aalborg


13. Edge Intelligence for Acoustic Sensing at a Massive Scale using Wireless IoT Devices

Wireless IoT devices are being deployed at a massive scale in the society and industry. Acoustic sensing is a relatively cheap way of sensing environmental phenomena, as sounds and vibrations are ubiquitous and highly informative. Nevertheless, putting an acoustic sensor on an IoT device has not been the most obvious use of IoT: Audio is usually associated with voice services and those services are not considered for IoT devices.

The objective of this project is to develop novel methods for acoustic sensing based on data collected from wireless IoT devices and processed by using deep learning algorithms at the edge/cloud. The research challenges are:

The potential of deep learning and acoustic sensing to improve the automation in manufacturing.
Use of edge intelligence to improve sensing performance and lifetime of IoT devices.
The effect of different audio codecs on the sensing performance, considering the sensor constraints, the monitored phenomenon and the IoT bandwidth.

Supervisors: Zheng-Hua Tan and Petar Popovski

Workplace: Campus Aalborg


14. Advanced Dithering Control Techniques for Switch Mode Systems

Dithering is employed in a wide variety of applications to overcome detrimental effects due to non-linearities. It works by purposefully introducing an additional noise or disturbance signal to a non-linear system with low-pass characteristics. A properly designed dither signal can linearize even severe non-linearities such as those found in switched systems. Dithering has been extensively used to achieve "super-resolution" in digital signal processing (most notable the Hubble space telescope), distortion-free re-quantisation between data types, or high-fidelity audio recording. Overall, the project aims to investigate and improve the use of dither in connection with feedback, to efficiently attenuate unwanted effects from switch mode systems. The work can focus on experimental results in the lab or be more oriented towards theory, depending on the background and interests of the PhD candidate. The project involves international collaboration with researchers from Norway and Australia.

Supervisors: John J. Leth and Jan Østergaard

Workplace: Campus Aalborg


15. Realization theory and model order reduction of deep recurrent neural networks

Recurrent neural networks (RNNs) are often the models of choice for deep learning algorithms. However, relatively little is known about 1) when and why these algorithms work, 2) how well the learned models will perform on new data, 3) how to find the minimal size of an RNN. In general, RNNs can be modelled such that they fall in the category of nonlinear dynamical systems, but for a widely used class of activation functions, RNNs can be consider as a subclass of piecewise-affine hybrid systems. Learning RNNs from training data can then be viewed as the problem of estimating parameters of dynamical systems. This has been studied for several decades in control theory and econometrics. This project aims to a) derive a realization theory of RNNs and apply this to analyse the correctness of learning algorithms, and b) apply model order reduction techniques to RNNs in order to find the minimal number of neurons. The work will be oriented towards theory and involve international collaboration.

Supervisors: John J. Leth, Zhang-Hua Tan and Rafael Wisniewski

Workplace: Campus Aalborg


16. Joint wireless communication and control loop design

In the Industry 4.0 vision, wireless systems are expected to replace wired industrial control networks. However, wireless communication and discrete control systems have been so far designed independently, resulting in suboptimal solutions that suffer from unnecessary utilization of radio resources and therefore poor scalability.

This PhD project aims at exploring the potential of a joint design of discrete control and wireless communication systems. The impact of wireless communication dynamics on the stability of time critical industrial control applications is to be analyzed, and novel tailored communication protocols as well as controller enhancements are to be designed. The main hypothesis of the project is that, such joint design can lead to a major reduction of the required radio resources for preserving closed loop stability, while limiting the controller complexity. Joint wireless communication and control loop design is identified as a major topic for 6G research.

Supervisors: Gilberto Berardinelli and Henrik Schiøler

Workplace: Campus Aalborg


17. Indoor localization of semi-static objects for IoT services

Ensuring accurate indoor localization remains a challenging and unsolved problem. Current development of IoT systems shows that indoor localization is a required feature, however, instead of highly accurate positioning, an IoT system often requires a correct association between an IoT sensor and a building plan. The project goal is to develop technology and algorithms for sensor-room association based on radio signals given only a single anchor point per room. Research challenges include analysis and measurements of radio links to extract observables suitable for determining sensor-room associations. Another important aspect of the project is Machine Learning. Existing AI frameworks can be exploited for reasoning whether two devices are in the same room, and this work can lead to development of new AI approaches. Algorithm design should be done under the constrains of IoT systems, including limitations in communication pattern, battery power and processing capabilities.

Supervisors: Tatiana Madsen, Jan H. Mikkelsen and Troels Pedersen

Workplace: Campus Aalborg


18. Distributed active perception for Internet of Robotic Things (IoRT) environments in smart hospitals

The Internet of Robotic is a novel paradigm that brings together autonomous robotic systems with the IoT vision of connected sensors and smart devices gathering and sharing information. This PhD project will investigate IoRT technologies to be used in smart hospitals, where fleets of robots perform routine and critical tasks, and replace humans when safety is a concern (e.g., in infectious disease wards). The PhD student will conduct research in different areas of robotics and wireless communications, from multi-robot planning and active perception to machine learning and IoT connectivity. The goal is to develop a robotic and communication solution for distributed perception of the robot, that intelligently fuses local sensor information with data received from distributed IoT devices and other robots. The project will be run in collaboration with the Smart Hospital Living Lab at the University of Melbourne, Australia, where the PhD student will have the chance to carry out an external stay.

Supervisors: Beatriz Soret and Letizia Marchegiani

Workplace: Campus Aalborg


19. Physical Layer Optimization of 6G Networks using Generative Adversarial Networks

Deep learning methods are expected to be one of the main novel technologies in future 6G mobile communication networks. They will be a stepping stone towards enabling, among others, the use of large intelligent surfaces, cell-free operation, and generally self-sustainable networks.  In this project, we explore the potential of generative adversarial networks (GANs) to solve fundamental 6G physical layer challenges such as: estimation and prediction of large-dimensional channels, interference modelling and prediction, or channel modelling for end-to-end physical layer optimization. We aim at developing novel, GAN-based algorithms that overcome classical methods and other machine learning solutions in terms of performance, computational complexity, or required amount of training data. Candidates with a keen interest and solid background in communications and statistical signal processing are encouraged to apply.

Supervisors: Carles Navarro Manchón, Morten Kolbæk and Elisabeth de Carvalho

Workplace: Campus Aalborg


20. Reconfigurable Intelligent Surfaces for Future Wireless Communication Systems

Recently, a new wireless channel concept has been proposed, which aims at adding special passive objects, reconfigurable intelligent surfaces (RIS), to the propagation environment that will intentionally modify the channel to provide better coverage and higher data throughput without the need of installing more base stations. This will be a benefit especially for 5G/6G systems on millimetre waves; additional benefits include wireless power transfer and improved security. A RIS system constitutes a new concept in wireless communication in many ways: in hardware design, in electromagnetic theory, communication theory, and optimization of a wireless network, and it will be necessary to integrate all these types of scientific expertise in the research of RIS. The candidate will study the underlying electromagnetic principles of RIS, propose higher layer models, optimize the performance of novel types of RIS, and evaluate the concepts by building prototypes and measuring their performance.

Supervisors: Ondrej Franek and Elisabeth De Carvalho

Workplace: Campus Aalborg


21. AI Acceleration at the Edge

More than 75% of enterprise data will be processed at the edge by 2025, while according to ARM, accelerating AI at the edge is the critical aspect of foreseen internetworking of trillions of IoT devices.

The PhD candidate will investigate and design a framework for computation acceleration at the edge that will enable low-power and IoT devices to perform complex Machine Learning and Artificial Intelligence operations. This AI-Edge framework will enable weak devices to harness the power of locally trained models for privacy-aware sensing at the edge. The project will advance the research by addressing challenges such as:

Edge Computing
Framework Heterogeneity with Containers
Intelligent Offloading Decisions
Hardware-Assisted Acceleration
Real-Time ML / AI Acceleration
Security and Privacy Considerations

The research will be conducted in an interdisciplinary collaboration between the CMI and the Automation and Control sections. The candidate with be affiliated with CMI in Copenhagen.

Supervisors: Cedomir Stefanovic, Sokol Kosta and Letizia Marchegiani

Workplace: Campus Copenhagen


22. Future Radio Spectrum Management

The aim of the PhD project is to give a techno-economic analysis of the future spectrum management strategies to contribute to decisions on management mechanisms to be implemented. The analysis will deal with specific ways of shaping these mechanisms technologically as well as in terms of governance and regulation. The focus will be on spectrum sharing and private networks. In the future, cognitive solutions based on advanced sensing and fast channel allocation may enable intelligent reuse of the incumbent spectrum by secondary users, leading to a more efficient spectrum utilization. This will require disruptive regulation mechanisms where secondary users may be allowed to camp over licensed spectrum where incumbents are inactive on a temporary and/or geographical basis. The PhD researcher will need a solid technological insight into wireless technologies and preferably also have some knowledge on frequency management. The PhD will be an MSc candidate, interested in the techno-economics of frequency management.

Supervisors: Anders Henten and Gilberto Berardinelli

Workplace: Campus Copenhagen


23. Wireless cabling – bringing space mission tactics down to earth

A key service of 5G wireless systems is Ultra-Reliable Communications (URC), aimed to replace cabled connections in critical applications demanding outages lower than 10-6. Presently, no experimental knowledge of wireless radio channels at URC-relevant regimes exists. Instead research is based on extrapolation of existing channel models.

Aim: turn the “impossible mission” of acquiring enough data for sufficient URC-relevant statistics into a “plausible” one by employing multi-domain sparse sampling in carefully tailored experiments inspired by space missions (factor in equipment reliability and diversity etc.).
Goal: characterize real URC-link conditions, i.e. give first “ground truths” and extract access system & device design impact.

We deal with rare events occurring at very low signal levels, so the candidate must have:

Strong mathematical base for stochastic modelling, rare events and big data.
Sufficient E.E. background for practical RF setup and signal estimation algorithms.

Supervisors: Patrick Eggers, Jesper Nielsen, Jimmy Jessen Nielsen and Søren Holdt Jensen

Workplace: Campus Aalborg


24. Turning technical KPIs for regional green energy into user incentives

The increasing focus on green energy in the society creates a need to present sustainable energy production, -flow and -consumption in easy-to-understand ways to consumers. The goal of this project will be to assert the hypothesis that generating the appropriate validation markers of buying/using green energy produced locally will in fact have a positive impact on end users’ consumption patterns. This is a challenging problem and requires an interdisciplinary approach involving technical depth related to e.g. communication and aggregation of measurement data, human/machine interfacing and user interactions studies. The system will provide useful metrics to express how green and regional the consumed energy is. These underlying metrics must be obtained and presented in a transparent and trustworthy manner to be accepted by the user. This PhD will contribute to a transparent and data driven approach for achieving a green and efficient energy infrastructure and providing user incentives.

Supervisors: Lars Bo Larsen and Rasmus L. Olsen

Workplace: Campus Aalborg




All news

See the list