Data and service placement on the Edge

A data placement technique for edge computing networks that places the data of a user according to their mobility.

User mobility measures at what rate a user is travelling through the edge servers of an edge environment.

The actions of the data placement strategy can perform are as follows:

• Place data according to the user's mobility class
• Retrieve data according to the user's mobility class

In order to evaluate the proposed technique we have developed a simulation as presented below. As far as the network topology is concerned, it regards a randomly generated network graph $G$, with 50 edge servers, and 100 users. Each node has a storage capacity which is randomly set and ranges from 500 MB to 2000 MB. Each user may have data whose size is randomly set within the range of 100 MB to 200 MB each. The graph edges (i.e., the links between the nodes) correspond to the delay (a.k.a., network cost) between the two connected nodes and is set randomly from 1 to 20 ms.

At every iteration we move the users according to their mobility class and make the proper arrangements with respect to the data placement. To execute this algorithm you can run the main.py file found in 'src/` as follows:

python3 src/main.py

If you are interested you may read more at our article:

@article{symvoulidis2023user,
  title={A User Mobility-based Data Placement Strategy in a Hybrid Cloud/Edge Environment using a Causal-aware Deep Learning Network},
  author={Symvoulidis, Chrysostomos and Kiourtis, Athanasios and Marinos, George and Tom-Ata, Jean-Didier Totow and Manias, George and Mavrogiorgou, Argyro and Kyriazis, Dimosthenis},
  journal={IEEE Transactions on Computers},
  year={2023},
  publisher={IEEE}
}