I'm afraid the following hints might not be to the point concerning the selection of par ameters. But to take a look might be worth while.

first is the Sleep Heart Health Study, allowing an extensive selection of multisample sleep stageing data, which is a real treasure. But GSR might be lacking.

second you might be succesful in searching the databases at physionet.org.

best,

Dieter

The Mahnob-HCI database for affect recognition. I worked with this database for several papers related to emotion recognition.

Check out https://physionet.org/physiobank/database/noneeg/

Gabriele Giordano you can use DEAP dataset from Queen Mary University of London https://www.eecs.qmul.ac.uk/mmv/datasets/deap/