The public defense of Rahul Nandkumar Gore’s licentiate thesis in Computer Science
The public defense of Rahul Nandkumar Gore’s licentiate thesis in Computer Science will take place at Mälardalen University, room Gamma (Västerås Campus) at 10.15 on November 10, 2021.
The title of the thesis is “Investigating Software-Based Clock Synchronization for Industrial Networks”.
The faculty examiner is Dr. Henrik Lönn, Volvo, and the examining committee consists of Dr. Henrik Lönn, Volvo, Prof. Christian Rohner, Uppsala University, Prof. Jerker Delsing, Luleå Technical University.
Reserve is Prof. Moris Benham, Mälardalen University.
The licentiate thesis has serial number 311.
Recent advances in the Internet of Things (IoT) are transforming businesses by leveraging digital technologies to change the existing business models and provide new revenue and value-producing opportunities. IoT achieves this by connecting physical objects that contain embedded technology over the Internet to interact with processes and personnel in an organization. For example, hundreds of sensors from a factory exchange data with a central control system over a communication network in an industrial domain. As a result, the data comes from a broader range of sources than ever before and helps factories fine-tune their maintenance cycles, maximize yield, and better understand the hidden costs of each decision.
The industrial IoT applications that monitor and control factory assets, require all the associated sensors and subsystems to perform their tasks orderly. In addition, the network devices need accurate time so that events can be logged and later analyzed correctly. Thus, a factory's proper and efficient operation depends on the correct sequencing of events and accurate data timestamping. Since factory devices clocks drift at different rates, they can show different times, resulting in erroneous timestamping and event sequencing. Such errors have a direct impact on data analyses and decision-making. To overcome this, time synchronization service aligns the network devices' clocks to ensure accurate timestamping and orderly event execution. Unsynchronized systems may lead to malfunctions such as false alarms, wrong decisions, and erroneous outcomes resulting in serious showstoppers for plant operations. Thus, an accurate and precise clock synchronization is essential for the correct and consistent operation of industrial systems.
Future industrial systems require a cost-effective, accurate, scalable, secured, easy to deploy, and maintain clock synchronization service for beneficing of IoT advances. Current software-based synchronization solutions are better suited to fit the future requirements than their hardware counterpart except for clock synchronization accuracy. With this motivation, the thesis focuses on software-based clock synchronization solutions to improve their accuracy by leveraging predictive software strategies. The clock synchronization service detects and minimizes the clock errors caused by internal hardware and communication disturbance by correcting calculated offset values towards predicted actual offset values. To this end, we proposed a new software-based clock synchronization method that performs better than state-of-the-art and state-of-practice solutions in our measurement dataset.