IoT research that focuses on people brought Radu-Casian Mihailescu to Malmö. In April, he joined IOTAP as post-doc in computer science. His main research interest is at the intersection of artificial intelligence and so-called smart grids.
Radu, why is there such a strong interest in IoT?
There is no doubt in my mind that the proliferation of intelligent objects in our surrounding environment will bring about fundamental changes to society. This poses a wide variety of challenges, but also opportunities for innovation.
First, I see huge opportunities for people and businesses to make more timely and better informed decisions based on a more quantifiable view of the world. This can mean anything from being more energy efficient, to finding new market opportunities, to smart healthcare products and so forth.
Second, while our brains are limited in their capability of perceiving the world through a limited number of senses, intelligent systems can overcome this limitation given the advent of low-cost connected sensors. We can thus interface with the environment with a different set of senses that can be virtually distributed across the world. Couple this with the systems’ ability to operate autonomously and reason about events in the physical world, it opens up a new realm of applications.
What is your research interest?
My interest is in the development of autonomous agents and multi-agent systems for socio-technical environments, so-called smart grids. My research is focused on decentralized and distributed coordination mechanisms, with an emphasis on stochastic, dynamic environments. In my approach, I combine a number of techniques from game theory, machine learning and combinatorial optimization.
Which IOTAP projects will you be involved in?
I will primarily be involved in CoSIS and a few other IOTAP projects where my focus will be on building intelligence into the system; intelligence in the sense of processing and harmonizing the data from heterogeneous IoT infrastructures, such that the collective intelligence of the system is gained from distributed collaboration and communication. The goal is to enhance the capabilities of the system and system components, through their ability to cooperate and being self-aware, while also considering the information security perspective.
With a background in artificial intelligence and multi-agent systems, I cannot help but notice that most IoT solutions are designed for very constrained domains, while the systems remain incapable to interpret data in a broader context. Such constrained domains are commonly handled by methods that would not work in a more contextually-rich scenario. Moreover, current solutions are geared towards collecting the data and using data scientists for performing the tedious task of making sense of the ever-increasing amount of available datapoints. With predictions of tens of billions of smart objects, this approach will become infeasible and the future of analytics in my view has to do with autonomously learning in real-time models from the stream of data, that can further be used for prediction, inference or anomaly detection, without having to actually store the entire datasets. It goes without saying that the flexibility and adaptability of self-managed systems will be paramount.
Why did you want to come to Malmö University?
With regard to the IOTAP center there are several aspects that seemed particularly appealing. Clearly, there are numerous interesting projects addressing timely topics in the area of intelligent management of IoT infrastructures. Moreover, having a research agenda explicitly focusing on the societal impact of these technologies and taking a people-centric approach is something that I can really relate to. Having a close collaboration with industry partners towards validating the undergoing research is again an important aspect, which I feel is strongly present also in the Swedish academia in general. Furthermore, the interdisciplinary context of the research makes it a very interesting environment.
» Radu-Casian Mihailescu received his PhD from Universidad Rey Juan Carlos, Spain.
» A recent paper focusing on ‘smart grids’ is Mihailescu, Klusch, and Ossowski’s eCoop: Privacy-Preserving Dynamic Coalition Formation for Power Regulation in the Smart Grid.