Test Dataset and Final Submission
The test dataset is available through the following link:
The test dataset will be released on the conference website here on July 10, 2021 (11:59 PM Pacific Time). Please check back here on that date to download the test set.
The test dataset will contain 13 sets of data with file names "test<x>.csv", where <x> is the sequence ID. Each sequence is under a different channel condition and contains 1000 transmission bits.
The participants need to decode every file and save the corresponding decoded bits into a CSV file with the name "test_bitsequence<x>.csv", where <x> is the corresponding sequence ID. The CSV format should be similar to the one in the training data (i.e., the bits are stored along the columns of the CSV file).
The final submission must include a 2-page abstract describing the algorithms and the methodology used, and a zip file containing all 13 decoded bit sequences in the test dataset. The format of the abstract should be similar to regular paper submission and the name of all participants must be included in the abstract. The final submission is through EDAS using the link:
https://edas.info/newPaper.php?c=28083
Data Competition
The object of the competition is to design and develop/train an algorithm that can detect the transmitted bits from the training dataset.
Possible solutions may build on classic algorithms (e.g., creating a new statistical model of the channel and use it to design maximum likelihood sequence detectors), or using purely data-driven machine-learning approaches, or a hybrid of these two.
Contest Details
For your convenience, a partial summary is presented below, but all competitors are strongly encouraged to read the complete contest details.
Dataset
The data is available through the following link:
Participation Instructions
To compete, teams should download
the dataset from the conference website and develop algorithms. On
July 10, 2021 (11:59 PM Pacific Time), a set of test data comprising of only the signal detected at the receiver (but no knowledge of the transmission bits) will be distributed.
The participating teams should run their algorithms on this test dataset and submit their estimated bit sequences no later than
July 11, 2021 (11:59 PM Pacific Time).
The results must be submitted as a CSV file containing the estimated bits, similar in format to the training data.
All competitors must submit a 2-page abstract, describing their algorithms and experiments on July 11, 2021 (11:59 PM Pacific Time) with their CSV files of the estimated bits, and present a poster at the conference.
The instructions for preparing the poster will be posted on the conference website.
Note that to be considered a valid participant in the competition, the competitors are expected to submit the 2-page abstract and attend the conference to present a poster, otherwise, they will be withdrawn from the competition.
The winners will be determined by the organizers, by evaluating and comparing the bit error rate (BER) over the entire test dataset.
Prizes
The competitors who are placed first, second, and third will receive certificates. In addition, the first-place and the second-place winners will receive a prize.
