The following topics are available for special assignments and theses carried out in Spring 2020.
The proliferation of resource-constrained mobile devices and smart objects in the Internet of Things has led to the generation of a large amount of data. Due to the recent advancements in Deep Learning (DL), services and applications based on Artificial Intelligence (AI) have become an enabler of smart cities, factories, intelligent transport systems and much more. DL models are often built from collected data (training), to enable the detection, classification, and prediction of future events (inference). Due to the limited computing resources at end devices, these models are often offloaded to powerful computing nodes such as cloud servers. However, it is difficult to satisfy latency, reliability, and bandwidth constraints while offloading data to cloud servers for training and inference of AI models. Thus, in recent years, AI services and tasks have been pushed closer to the end users – to the fog – to meet these requirements. The main objective of this project is to implement a DNN inference offloading framework over fog networks developed in our research group and to evaluate its performance.
Smartphones are increasingly being equipped with depth sensors, which measure depth directly and thus can be used in 3D reconstructions as well as volumetric streaming. In fact, recently, Samsung has recently released 3D scanning app that runs on their latest Samsung Galaxy Note and S10 5G smartphones. This application creates small-scale 3D models on the phone. The rise of 5G and edge computing will play a key role in enabling applications that can process large-scale depth data offloaded from smartphones. For instance, tele-presence applications can directly use depth data to reconstruct 3D models of people to enable new and immersive forms of communication. Another promising application is the real-time streaming of the 3D environment in which a user is present. The goal of the thesis is to evaluate how to offload depth data from smartphones and to build a simple application that relies on such data generated by smartphones. The thesis will investigate the wireless bandwidth requirements of such applications and propose new schemes to efficiently offload such data.
The Internet of Things (IoT) has motivated employing different technologies to satisfy diverse application-specific requirements. As a result, heterogeneous networks have to coexist in the same physical space and share the same frequency band, resulting in throughput degradation. Several works in the literature have addressed the problem of detecting interfering technologies; however, they focus on one or a few technologies. Moreover, the use of Machine Learning (ML) techniques as a tool to classify the signals and detect their native technology has not been carefully studied, even though it is recognized as a promising approach. The goal of this project is to implement a platform-independent ML algorithm to classify foreign interfering technologies, such as WiFi, ZigBee, Bluetooth, and so on. The project involves carrying out both simulations and experiments with software-defined radios to show the effectiveness of the algorithm.
Pervasive displays are widely employed in public spaces to convey information to users. For instance, they are deployed in airports to provide information about flights status and in malls to show offers available at the stores therein. However, displays tend to be ignored by users if they are found to be neither informative nor interesting. Thus, a key challenge is the design of solutions that provide interesting content to users with limited time and attention. In particular, content selection is particularly challenging for tiled pervasive displays that show multiple content items at the same time. The goal of this project is to extend an existing approach developed in our group which relies on the information foraging theory for adaptive selection of display content based on audience data. The project involves analysis of data collected from a depth camera as well as the evaluation of the improved solution under real settings.
The software development process is evolving. Serverless computing (or function-as-a-service) is an emerging paradigm, wherein software developers can deploy stateless functions without worrying about the underlying infrastructure. These functions are executed only when triggered; for instance, as part of a processing pipeline that starts with the generation of sensory data. Particularly important applications include those enabled by the Internet of Things (IoT). The goal of this project is to examine recent serverless frameworks, especially in the context of edge and fog computing. In particular, the student is expected to review the related state-of-the-art on serverless computing as well as describe the main benefits and challenges, with special reference to the IoT. The project involves a performance evaluation of severless platforms (including Knative and Apache OpenWhisk) in the context of the IoT.
Learning about new technologies is challenging if not adequately supported by appealing hands-on activities. A compelling approach is represented by interactive labs: an online environment made available to users as a web application over the Internet. Users can then follow a tutorial and carry out the related tasks directly in the browser, without the need to install any software locally on their computer. This is particularly beneficial for many cloud-native technologies (such as Docker and Kubernetes), the installation and setup of which could be very cumbersome in the first place. The goal of this thesis is to build an online learning environment for interactive labs as a web application, similar to Katacoda. The web application enables users to carry out interactive tutorials directly in their browser and allows an administrator (e.g., a teacher) to to keep track of the progress / outcome of the labs. The thesis work includes the realization of a prototype with at least some of the features of a complete system.
Pairing is the process of establishing a connection between two personal mobile devices. Even though such a process is intuitively very simple, achieving a secure pairing is challenging due to several possible attacks and usability-related issues. The goal of this project is to design and implement a new scheme to securely pair a smartwatch and a mobile device (e.g., a mobile phone) through gestures involving drawing, supported by accelerometer data. For instance, the user can pair the smartwatch by drawing with a finger on the screen of the mobile device.
I am always looking for motivated students willing to get a PhD. If you want to learn more about PhD studies at the Department of Computer Science of Aalto University, refer to the Doctoral Program in Science and the how to apply page.
Aalto University is a young but internationally recognized university, especially in the field of computer science. The main campus of Aalto University is located in Otaniemi, the largest ICT and business hub in Northern Europe, and hosts the Helsinki node of EIT Digital, a knowledge and innovation community of the European Institute of Innovation and Technology. Besides, Finland is one of the most livable countries in the world.
I am offering PhD positions in the broad area of network systems. They span from wireless communication to mobile computing and distributed systems, including the Internet of Things. The actual topics offered vary over time and depend on many factors, including my current research interests and funded research projects. You can have a look at my recent publications and research projects to get an idea on how they could look like.
I participate in the Helsinki Doctoral School in ICT (HICT) under the Networks, Networked Systems and Services track. If you would like to do a PhD with me, apply to one of the HICT calls (there are usually two per year) and provide my name as supervisor. Do not email your application material to me but follow the instructions on the HICT call.