Extending Linux Kernel Paging Mechanism for Hybrid Main Memory Management using Reinforcement Learning

We are seeking a motivated and research-oriented student for a research project to develop a functional implementation of an RL-based paging mechanism within the Linux kernel, with the objective of intelligently allocating pages between dynamic random-access memory (DRAM) and non-volatile memory (NVM). This research project offers challenges and opportunities to work at the intersection of kernel-level development, memory management, and reinforcement learning. The algorithm has been implemented and evaluated in a simulated environment and will be used as a basis for the implementation on the kernel level for a real system evaluation. 

Skills and Qualifications:

• Strong programming skills in C (knowledge of Assembly and/or C++ is a plus) 
• Experience in kernel-level development and understanding of OS design principles 
• Familiarity with computer architectures and memory management concepts 
• Good problem-solving and analytical skills 
• Previous experience or coursework in reinforcement learning or related fields (preferred) 
• Ability to work independently and collaborate effectively with a research team 

Responsibilities: 

• Conduct a comprehensive literature review on paging mechanisms for homogenous and heterogenous memory systems 
• Collaborate with the research team to integrate the RL-based paging mechanism into the Linux kernel source code 
• Address implementation challenges, such as optimizing Q-table placement and minimizing implementation overhead 
• Conduct performance evaluations and comparative analysis of the built-in and RL-based paging mechanisms using memory-intensive workloads  
• Document and present research findings in the form of a thesis or a research paper 

Benefits: 

• Opportunity to contribute to cutting-edge research in memory management 
• Hands-on experience in kernel-level development and integrating RL algorithms into complex software systems 
• Collaboration with a team of experienced researchers and industry experts 
• Possibility of publication and recognition for outstanding contributions 

Type of Project: Internship 

Master's degree: Master of Engineering Technology; Master of Engineering Science; Master of Science 

Master program: Computer Science; Electrotechnics/Electrical Engineering 

Duration:  6 - 9 months 

For more information or application, please contact one of the supervising scientists:

• Emil Karimov (emil.karimov@imec.be)
• Tommaso Marinelli (tommaso.marinelli@imec.be) 
• Saeideh Alinezhad Chamazcoti (saeideh.alin@imec.be)

Only for self-supporting students.