sensors-software
sensors-software copied to clipboard
opendata-stuttgart
Acces Point mode: wifi password transmitted unencrypted? Maybe a security problem
A friend and me, we are thinking about building a Feinstaubsensor, therefore I had a short look at the project and the code.
In the first setup, when you have to enter the settings for your home wifi, is this information transmitted encrypted?
On first sight it seems to me that it is not.
In this line the AccessPoint is established with a password stored in cfg::fs_pwd however, as this line shows, cfg::fs_pwd is an empty string.
Therefore, it seems to me that an attacker that "listens" to the communication between me and the ESP8266 while entering the settings for my home wifi could get access to my network like this.
Maybe I missed some other security measures you have taken. If not, an easy solution would be to set cfg::fs_pwd to something else than an empty string.
I've also noticed this issue: https://github.com/opendata-stuttgart/sensors-software/issues/128 however, I couldn't see if this particular problem was already solved with the measures you were taking there.
The password is transmitted unencrypted on send. What can we do?
- using HTTPS: this would need a certificate, but self-signed certs are throwing a warning (confusing most users)
- using WPA2: we need to set a random password. But how should we tell this password to the user? Not all users are able to read the console output. If you think you will have to change the password regularly, then you can set an own password for the sensor in AP mode
Normally the wifi password should be transmitted only once unencrypted at the moment of saving (and only if changed). In all cases where a 'password' is shown in the input fields only a string with asterisks is transmitted. So someone needs to capture the traffic in exactly this timespan where you configure the sensor the first time.
If you want to avoid the the unencrypted transmission you could compile and flash a modified version of the sensor software where you can predefine your wifi password.
Thank you for your fast response.
Personally, I would go for method 2 (WPA2), because it is very easy to implement.
However, I doubt that it is really necessary to set a random password. For sure: setting a random password would be the most secure way to deal with this problem.
However, even setting a static password like "feinStaub123" that is mentioned in the instructions would be much better than the actual solution. In my opinion this would even be secure enough, as long as the SSID does not reveal what project is currently installed on the ESP8266.
After all, there are thousands of different ESP projects not including unpublished "homebrew" projects. As each project uses (or at least should use) a different password, even if the password is static it would be very hard for an attacker to find the correct decryption phrase.
This solution is also used here: https://github.com/sidoh/esp8266_milight_hub/blob/master/src/main.cpp#L289
As long as nobody has a better practicable idea, I would strongly suggest to use a static password for the access point instead of not using any password at all.
Just set up two sensor-systems and realized this issues exists right after. I totally agree with speendo. Just "Password" as a password would be a big improvement as it prevents sniffing of the actual wifi-password while it is being transmitted.
This still leaves the issue that someone could be able to reconfigure the sensor, but I don't really see the security-issue with that, as it runs with default configuration at that point
I've made up a pull request to overcome this issue. Only a minor changes. However, the manual(s) would have to be adapted...
Please consider accepting the pull request as this problem could possible cause some harm.
IMO, you should always consider what is exactly the problem. Is it that, with the WiFi password, anyone can access the internet over your access point? Or is it that other people can snoop on YOUR network if they have the WiFi password?
@bertrik I think the question is valid in general.
However, in the particular case I would say that the exact problem is, that your wifi password can not be considered secret, when you transfer it on an unencrypted connection.
A potential attacker could do several bad things once they have your wifi password INCLUDING but not LIMITED to accessing your internet connection and snooping your network.
As the fix of this problem is not difficult and I reported this issue more than three years ago, I think it would be about time to implement a solution, regardless of what the worst kind of attack would look like.