Improving ZeroMQ based communication in Concore

[Afreen-Kazi-1]

Hello @shivangvijay, @Rahuljagwani and @mayureshkothare.

I’m Afreen Kazi, a sophomore pursuing a BTech in Computer Engineering at Veermata Jijabai Technological Institute, Mumbai, India. I’m passionate about AI, Machine Learning, and Computer Vision and am eager to contribute to Alaska for GSOC'25.

I have experience with Deep Learning (TensorFlow, Keras), OpenCV, ML algorithms, and Unity, and have previously worked on a Virtual Try-On project that generates realistic clothing try-ons for both images and videos. You can check it out here

I've gone through the concore codebase, and I understand that it facilitates modular simulation for closed-loop neuromodulation control systems using file-based and shared-memory-based communication. Previously, ZeroMQ communication has been implemented for concore using Osparc-control which led to issues like ZeroMQ hyper edges not being visible which in turn made the study representation look like a forest instead of a directed hypergraph and that meant running each hypergraph in the study separately. I’m particularly interested in

For improving concore's ZeroMQ-based communication while ensuring seamless integration with the concore editor I'll have to:

• Ensure ZeroMQ edges are properly represented in the concore editor with the hexadecimal port-based naming scheme
• Modify DHGWorkflow visualization to maintain a single directed hypergraph by extending concore's graph representation
• Update graph parsing algorithms to treat ZeroMQ edges the same way as file-sharing and shared-memory edges.
• Benchmarking performance improvements over existing file-sharing mechanisms to validate efficiency and usability.

Also, I would like to propose implementation of a mechanism where users can mix all three communication methods (file-sharing, shared-memory and ZeroMQ-based) for efficiency, reducing overhead for local studies while enabling distributed execution when needed. and allowing flexible configurations based on system requirements.

I would also like to suggest implementing automated port management for ZeroMQ hexadecimal port-based labels to prevent port conflicts and simplify study configurations by dynamically managing ZeroMQ ports instead of requiring users to manually assign them. I would like to hear your thoughts on this!

Would it be possible for me to email you a detailed draft proposal for feedback? I’d love to refine my approach based on your insights.

Thanks,
Afreen Kazi

[shivangvijay]

Hello @Afreen-Kazi-1,

Thank you for the introduction; your profile looks great. Please note that we do not accept any email communications until the accepted GSoC contributor projects are announced. Kindly share all your questions and thoughts here.

[Afreen-Kazi-1]

Alright @shivangvijay.
I appreciate your response! I'd love to hear your thoughts on the approach I mentioned as well as the suggestions particularly regarding:

• Hybrid Communication Support – Allowing users to mix file-sharing, shared-memory, and ZeroMQ-based communication within a single study for greater flexibility and efficiency.
• Automated Port Management – Dynamically assigning ZeroMQ ports to avoid conflicts and simplify study configurations.

Would these align with the vision for concore? Are there any specific challenges or considerations I should account for while implementing them? Your insights would be really valuable in refining my approach!

Looking forward to your feedback. Thanks!

[Afreen-Kazi-1]

@Rahuljagwani, @shivangvijay, @mayureshkothare
Are there any tasks or PRs I need to complete for applying for GSOC? Can you please guide me about the procedure at ALASKA.
Thankyou.