I think it might be worth summarising where we now are in terms of test protocol.
Revised hypothesis: PS can determine whether the WiFi is on, and operating at a level greater than Y for Z minutes, from his physical symptoms
Revised test protocol
Equipment required: wifi modem, EMF meter and stop watch
Values X, Y and Z to be specified by PS before the first trial.
1. On a day when PS feels up to it, he goes into a room in his house separated by at least one room from the room containing the wifi modem (off at start of trial), whilst his wife remains in the latter room
2. At a predetermined time his wife tosses a coin
a) if the coin is heads she turns the WiFi on, writes ON on a piece of paper, seals it in a numbered envelope and switches the meter on
b) if the coin is tails she writes OFF on a piece of paper and seals it in a numbered envelope
3. For the next X minutes:
a) PS monitors his state of health
b) If the modem is on, PS's wife monitors the EMF meter and starts a stop watch whenever it shows a value above Y, stopping it when it shows a value lower than Y. If the modem is off, she remains in the room and passes the time quietly in any way she sees fit.
4. After X minutes PS writes either ON or OFF on a piece of paper and seals it in a numbered envelope, then rejoins his wife
5. If the wifi was off, or it was on and the stop watch shows that the wifi modem was above level Y for a minimum of Z minutes, the two envelopes are added to the pile of successful trials. If not, both envelopes are discarded
6. Repeat steps 1 to 5 until ten successful trials have been conducted
7. If at least eight of the ten pairs of envelopes contain the same word, the hypothesis is tentatively accepted and a more formal, witnessed test (perhaps adjudicated by CfIIG) is arranged. If not, the hypothesis is rejected.
Note 1: I've specified a one room separation because in most modern houses it's possible to hear sounds like equipment and stop watches being switched on and off through the walls. If PS feels he needs to be closer to the modem than that, some way of masking sounds (e.g. having a radio playing in one or both of the rooms) might be substituted. Alternatively his wife could monitor the wifi level from a different room.
Note 2: In a formal test PS should be unaware if he has guessed correctly in any individual trial until the experiment is complete, but in this informal test scenario that is not possible.