SAWEC

This is the implementation of the paper SAWEC: Sensing-Assisted Wireless Edge Computing and Integrated Sensing and Communication for Efficient Edge Computing. The repository shares both the datasets and the source code of SAWEC.

If you find the project useful and you use this code, please cite our papers:

@inproceedings{haque2024integrated,
  title={Integrated Sensing and Communication for Efficient Edge Computing},
  author={Haque, Khandaker Foysal and Meneghello, Francesca and Restuccia, Francesco},
  booktitle={2024 20th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)},
  pages={611--614},
  year={2024},
  organization={IEEE}
}

and

@article{haque2024sawec,
  title={SAWEC: Sensing-Assisted Wireless Edge Computing},
  author={Haque, Khandaker Foysal and Meneghello, Francesca and Karim, Md Ebtidaul and Restuccia, Francesco},
  journal={arXiv preprint arXiv:2402.10021},
  year={2024}
}

Firstly clone the repository with git clone git@github.com:kfoysalhaque/SAWEC.git

Facial Recognition Test (Preliminary Test)

(I) Go into the test directory with cd Facial_Recognition and download the trained_model and the dataset within the directory.

(II) Unzip the dataset and delete the zip file with sudo unzip Test_Data.zip && sudo rm -rf Test_Data.zip.

(III) Create different compression quality dataset

python compression.py <'Compression Quality'>  && python preprocess_compressed.py <'Compression Quality'>

Example: python compression.py 10 && python preprocess_compressed.py 10 compression quality of the image ranges from 0-100, higher the value, the better the image

(IV) Create different reshaped (downsized) dataset

python preprocess_image_downsize.py <'image_resolution'>

Example: python preprocess_image_downsize.py 256

(V) Train the model with 1024x1024 image shapes: python main.py

(VI) Evaluate the performance with different image compression and image downsize ratio

python evaluate.py <'test_name'> <'compression_quality_or_image_resolution'>

Example:

python evaluate.py compression 10

python evaluate.py downsize 256

test name- "compression" / "downsize" and compression quality==> (0, 10, 25, 50, 75, 100), the higher the value, the better the image quality. For the image downsize test, input image resolution==> (1024/512/256/128/64/32/16)

Sensing Assisted Wireless Edge Computing

(I) Please follow our mD-Track implementation for localization. However, you might perform localization with any other algorithms, resulting in different performances.
(II) Please download and unzip the localization information and the corresponding 10K frames.

sudo unzip Stitched_Video.zip && sudo rm -rf Stitched_Video.zip 

(III) Extract the partial frames (10K resolution) using the localization information

cd Partial_Frame

python main.py <'environment'>

Example: python main.py Classroom1

Name of the environment could be, "Classroom1" or "Anechoic1"

The partial frames will be in the newly created directory "Partial_Frames_10K".

You can downsize or compress the partial Frames by executing python downsize_or_compress.py <'environment'> <'conversion_type'> <'ratio'>

Example: python downsize_or_compress.py Anechoic1 downsize 256

(IV) Move into the directory Segmentation_Performance for evaluating object detection and segmentation task

cd ../Segmentation_Performance

Then fetch the partial frames with python fetch_images.py Please set the source and target directory path accordingly

(V) download the YOLOV8 models from here and unzip them with unzip yolov8_trained_weights.zip && rm -rf yolov8_trained_weights.zip

Execute fetch_images.py with the correct directories to fetch the partial frames. Execute evaluate.py to analyze the performances. Please change the yaml files accordingly.