I have written a code in network simulator 3, which is based on C++. It is simulating a new protocol within a 5G network. I have another code in Google Colab, by using one of the well-known deep learning libraries called TensorFlow. Now I have my code in network simulator 3 and my trained neural network in Google Colab. I want to send some parameters from the Network simulator 3 (C++) to Google Colab and then return the output based on the neural network to the network simulator 3, and make some decisions based on the output of the neural network. How can I establish a connection between these two platforms?
Thanks in advance
Have a look at https://au.mathworks.com/videos/using-matlab-with-python-1591216182793.html
ALSO ::: https://au.mathworks.com/help/matlab/matlab_external/connect-python-to-running-matlab-session.html
https://realpython.com/python-bindings-overview/
https://wiki.python.org/moin/IntegratingPythonWithOtherLanguages
you can run the code in the python shell using cython package.
https://cython.org/
Python is best as a Glue/Script machine, use it at the highest abstraction level to call the simulator, gather the results, run your Colab code and then call the simulator again with the data from Colab. That's probably the best way to get fully automated without risking adding bugs and complexity that will stop things dead.
If you need an all-in-one solution then don't forget that once you have the graph you can use the C++ version of the tensorflow API to run it directly (and thus have a single code-base).
However, as mentioned above, a process connection such as a socket or a storage / file-system bridge (like an RDBMS or just a single file) can be incredibly easy to setup and get going (before making something more substantial). It also opens the door for distributed versions when you scale up.
- Badges
- Science topic