First, to answer your titular question:
Is it possible?
Likely yes. Otherwise, all PhD students at this university would fail, wouldn't they :) ? Let me go over the rest of your question one by one:
It seems that most universities in Europe require an outline of the planned dissertation at the application stage.
Most seems a bit extreme. I know that this is how it works in some universities, but it certainly did not work like that in all places I worked in.
I think even choosing the title of a dissertation needs a lot of dialogue between the student and his supervisor. It also requires a thorough investigation on the state of art in the targeted area.
Correct. At my current university, people hand in their proposals during their second year usually.
Is such proposal the definite proposal or it may totally be changed after the admission?
It is almost certainly not a very definite plan, but whether it can totally change I am not sure. For instance, I would assume if it changed so much that it started to fall out of the area of expertise of your advisor, I would imagine things would get tricky.
Can you provide some tips for writing such proposal draft? How much time you think I should devote for such plan (at least)?
Rsearch the state of the art in the field you are interested in. Take a few days minimum to browse over the keywords of the papers of the top conferences in the field. Find out which professor at your university publishes in these top conferences (if there is nobody, this university may be a bad match for your field of interest), and see what the typical keywords and style of his work are.
Think about ~3 coarse-grained research questions that you think are not answered yet by existing work. You probably already needed to define a research question for e.g., your master's thesis. Make sure that the scope is a bit broader now for a PhD - you don't want research questions that are basically answerable within one paper in a few months of work (Bad: "Q1: is it possible to apply algorithm A to problem B?"). On the other hand, you do not want to be too general either (Bad: "Q2: how can security be introduced in service-oriented systems?" - this one is a real-life example).