Smart homes handle data locally
How many connected devices are there in the world? No one knows for sure, but according to a forecast, we might reach 50 billion in a couple of years. This means that enormous amounts of data have to be generated, managed and stored. Today’s cloud solutions won’t be enough, argues Ramin Firouzi.
For most of us, “the cloud” is a fuzzy concept. We don’t know where the cloud storage happens or how secure it is. “My photos are somewhere in the cloud”, we might say, perhaps gesturing towards the sky. But the cloud is actually more grounded than it sounds.
“Our data is stored on servers in data centers that are owned by big tech companies”, explains Ramin Firouzi, who holds a PhD in Computer and Systems Sciences.
He has spent the past few years studying an alternative to managing and storing data in the cloud, namely edge computing. As the name suggests, data is processed closer to the user, on local devices and servers.
The number of connected things is increasing exponentially
The development of Internet of Things (IoT) is an important factor. We’re buying more and more sensor-equipped gadgets—smart refrigerators, robot vacuums, connected toasters, and watches that monitor every step we take.
“Currently, the number of connected things is increasing exponentially, no one knows where it will end. We’re talking about possibly having 50 billion connected devices by 2025. Technological development is so rapid that all forecasts become uncertain”, says Ramin Firouzi.
Cloud and edge must be combined
To handle the enormous amounts of data, new approaches are required. Edge computing isn’t really new—the technology has existed for a decade. But perhaps it’s not until now that it has become apparent that cloud technology isn’t sufficient.
“It’s not cloud or edge – we need both. The technologies must be combined. I’m interested in how edge technology can be used effectively to improve both computation and communication.”
Firouzi sees two major problems with cloud computing. One concerns privacy and security. Even though Sweden has fairly strict rules for handling personal data, there’s a looming risk of cloud servers being hacked and sensitive material leaking.
The other problem is the bottleneck that arises when sending gigantic amounts of data to one place. It creates a delay that, in some cases, can be critical.
“When computation is made closer to the user, it’s much faster. For example, if you have a smartwatch that detects your heart rate and other values, it can alert immediately if you’re about to have a heart attack. And in that situation, seconds can be crucial”, says Ramin Firouzi.
Safer to store data locally
It’s unnecessary to send all of your health data to the cloud, he argues. It’s enough to send certain parameters which will allow calculations to be made on aggregated data in the cloud. That would also be safer.
Firouzi mentions future smart homes as an example of how edge and cloud computation can be combined. We want heating, ventilation, sanitation and other functions to be efficient and tailored to how we live our lives. At the same time, for security reasons, we don’t want detailed data about when we’re at home or at work to be uploaded to the cloud.
“If your data stays in your house – on your local server – the risk of leaks decreases significantly.”
When we handle our data locally and only send certain parameters to the cloud, they can be processed along with parameters from other households. The results of the calculations can then be sent back to our local devices to make our homes even smarter.
“Machine learning is about computers learning human behavior. By training models on large amounts of real data, the models get better at making predictions. It could be, for example, to predict at what time in the day we need hot water to take a shower”, says Ramin Firouzi.
Critical applications in healthcare
In his thesis, he develops a framework for so-called distributed intelligence, where data processing occurs on many smaller devices instead of in one place. Applications exist in many different areas, particularly in healthcare, where more efficient transmission and processing of data can save lives.
“The development of communication networks is a strong contributing factor to the sharp increase in the number of connected devices. I have explored how edge technology can be integrated into 5G and 6G so that IoT systems become more efficient. It’s absolutely crucial for the development of the field”.
I have explored how edge technology can be integrated into 5G and 6G
Next step for Ramin Firouzi is to take the knowledge gained as a PhD student at DSV into the business world. As a data scientist, he will work on integrating edge and cloud technologies. But he’s not closing the door to academia.
“In the future, I hope to be able to continue my academic work. Should an opportunity arise, I would gladly return to DSV. I moved from Iran to Sweden when I was admitted as a doctoral student, and everything here was new to me. I’ve learned so much from both colleagues and students”, says Ramin Firouzi.
About the research
Ramin Firouzi successfully defended his PhD thesis on October 13, 2023 at the Department of Computer and Systems Sciences (DSV), Stockholm University.
The title of the thesis is “Distributed Intelligence for IoT Systems using Edge Computing”.
Ramin Firouzi’s main supervisor is Rahim Rahmani, DSV, supervisor is Thashmee Karunaratne, DSV.
Sadok Ben Yahia, University of Southern Denmark, was the external reviewer at the defence.
Text: Åse Karlén
Last updated: November 24, 2023
Source: Department of Computer and Systems Sciences, DSV