Theres a url, say peepee.com. So far this is the routing portion of the url that says how to find the web server, basically saying "ask .com how to find peepee", and that gives us the ip address of the server.

Everything that comes after that, is information for the server itself. So to navigate to a resource, say poopoo, that lives on the server, they would navigate to peepee.com/poopoo.

But sometimes you want to navigate to that resource and also communicate some bit of information to the server, say a login token so the server knows who is accessing that resource. This is communicated via a URL parameter, and looks like ?userid=abcd1234, or in the full url: peepee.com/poopoo?userid=abcd1234. So the user is still accessing the same resource, but has provided additional metadata to the server.

These parameters can be abused to identify who knows who and who communicates with who by attaching a tracking id parameter to the URL, so when you share a link it includes that tracking parameter and anyone who clicks on it, well now the server knows that the originator of the tracking ID (well, the first person to be assigned it) shared it with this other person. This can be combined with other collected info to build a map and social graph of actual people, e.g. we know dave is at this ip, and jane is at this other ip, and we put a tracking parameter in daves url and we saw jane use that same tracking parameter in her url, so we know that dave shared this url with jane.

So to answer your question, a canonical link is a link to a resource without the unneeded url parameters.