BCDC thesis: What you should know about edge computing and how to make it workKirjoittaja Kaisa Ikonen News
BCDC-researcher Nitinder Mohan’s thesis defence was held at the University of Helsinki on November the 8th 2019. In this BCDC’s interview Nitinder tells us about the content of his doctoral research, the research methods, and the results and the applicability of his results in a multidimensional manner. In addition, he discusses the importance of his research in achieving a wider target and goals; that is, for future computing development, enabling next generation applications and energy systems’ disruption.
Please, tell us who are you, and where did you conduct your doctoral thesis?
My name is Nitinder Mohan and I conducted my doctoral studies and research in the Department of Computer Science, University of Helsinki and as part of the BCDC Cloud Team.
What is the topic of your thesis and what are the main contents of it?
My thesis is titled “Edge Computing Platforms and Protocols”. It provides a comprehensive collection of novel solutions for integrating edge computing in existing cloud infrastructure.
In recent years, interest in applications such as Internet-of-Things, augmented reality, self-driving vehicles, smart energy etc. has risen dramatically and researchers are finding ways to support the operation of such applications. However, these applications are extremely time-critical as they operate in real-time. Existing cloud computing environment, which utilizes datacenters deployed in far off locations, cannot support the computations for such applications due to considerable network delays for transferring all relevant data.
Edge computing solves this problem as it aims to utilize existing (laptops, smartphones etc.) or specially designed compute servers in the vicinity of the users and sensors. Such servers, although lack in compute capacity compared to datacenters, significantly reduce the network latency for computations.
As the field is still in its infancy, my research tackles several issues that hinder the adoption of edge computing. A few such challenges are optimal deployment locations, distributed data/computation handling, optimal utilization of wireless network, along with many others.
What kind of new information and new results does your doctoral research offer?
My thesis identifies several challenges that obstruct the effective operation of edge clouds. The dissertation provides the researchers in edge computing an “edge computing service architecture” that organizes different areas of research opportunities for integrating edge computing with cloud computing.
The thesis provides resolutions for a few highlighted challenges. For example, through extensive measurements, my colleagues and I have found that currently available wireless and cellular technologies, e.g. WiFi and 4G, cannot replicate the reliability and performance offered by datacenter networks. Therefore, we proposed QAware that allows edge servers to simultaneously utilize more than one network connection – i.e. a combination of WiFi and 4G – using Multipath TCP optimally. There are several solutions that this thesis provides for the practical utilization of edge servers.
What is the role of your doctoral research and its results in relation…
…to the research tasks, targets and goals of the BCDC project and the Cloud Team?
The focus of the BCDC Cloud team is to bring distributed cloud services and computations closer to users and the edge. The advantages of such a distribution is twofold. Firstly, deploying network services closer to the users allows applications to react faster and tailor to local requirements. A prime example is an autonomous vehicle driving within the city and offloading all camera and sensor data to the nearest edge server for computations. Secondly, it allows effective utilization of green energy for doing computations. For example, users can explicitly specify to use only those edge servers for their tasks which are being powered by renewable energy.
This thesis achieves both of these objectives of the BCDC Cloud team. Firstly, it tackles a variety of operational and software challenges for distributing cloud computations and services at the edge. Secondly, it provides eLPCF – a task deployment framework for edge clouds – which minimizes the overall energy for carrying out computations in edge clouds.
…to the targets and goals of Strategic research?
BCDC is part of the Academy of Finland’s Strategic Research Council funded Disruptive Technologies and Changing Institutions (2015–2021) research program. How does your research relate to the goals of this research program?
In my thesis, I have proposed several protocols and platforms that enable the effective deployment, operation and utilization of edge computing. Not only is edge computing a disruptive technology, but it also paves the way for other emerging disruptive applications – e.g. for smart energy, smart factories and autonomous vehicles – by supporting their computation. In other words, it gives answers to the question asked by the Strategic Research Council: “In order to make the best possible use of that particular disruptive technology, what changes are required in human activity, institutions and operational methods?”
… to the development of the future computing?
Cloud computing has created a radical shift in expanding the reach of application usage and has emerged as a de-facto method to provide low-cost and highly scalable computing services to its users. However, novel applications, such as Internet-of-Things, augmented-reality and autonomous vehicles, push the capabilities of cloud computing as it can no longer support the ultra-low network latencies required by such applications.
Edge computing brings a radical change in cloud computing services and envisions a future where computations will take place close to users and applications instead of remote data centers. Smartphones, tablets, routers, base stations, and many more devices will provide computational capacity enough to support the requirements of emerging applications. “Computation-as-a-Service” will become pronounced and will be detached from the location, type, and capacity of the server. This thesis provides solutions that bring computing research one step closer to this vision.
Which actors etc. can benefit from the results of your thesis?
I hope that my thesis can help existing cloud and cellular network providers who want to deploy and utilize edge servers to better support emerging applications. The dissertation provides several solutions that integrate edge servers within existing cloud infrastructure and technologies, which would make the transition from centralized compute farms to distributed computing uncomplicated.
What method did you use in your research and what do you think about this method’s role for your research? Was this a successful choice?
As this thesis offers a collection of protocols and platforms for every layer of the cloud service model, I used a combination of mathematical proofs, simulations and running code to evaluate my proposed solutions. For example, for assessing the correctness of our task deployment framework, LPCF, we relied on Python-based simulations. These simulations allowed us to also test for corner cases that may not present themselves as commonly in real deployments. On the other hand, we relied on real-system testing – including implementing our solution in the Linux kernel – to measure the performance and behavior of QAware, which allow edge servers to utilize multiple network access technologies (e.g. WiFi + 4G) simultaneously.
Using a mixture of different methods is quite common in systems research. It allows for thorough testing of your solutions, providing confidence that the protocol does not fail to handle extreme cases. So Yes, I would consider it a successful choice.
Lastly: How is it going on now? Where are you working and what kind of issues do you work with?
I will join as a postdoc at the Technical University Munich, Germany from the beginning of the year 2020. I will be working with Professor Dr. Jörg Ott in his Connected Mobility group. The group is working on edge and cloudless computing in the context of user and network service mobility.
I am hoping to apply many of the insights I gained during my Ph.D. thesis process to the research questions I will tackle in Munich.
BCDC congratulates Nitinder for successfully defending his doctoral dissertation and wishes all the best for his current and future works!
Photo below: iStock