It is entirely possible to find items with known addresses across an ad hoc network.
If you want to think of it in terms of a global database (which is not an apt mental model), then the collective topological knowledge of all the nodes in your network constitute the database, and the problem reduces to remote information queries.
If you have communicated with your identity, then the information of your identity's location is cached by those nodes which conveyed the information.
I could go on for hours about it, but the basic is: it's possible, it's mathematically feasible, and it will work more or less as described.
If you want to think of it in terms of a global database (which is not an apt mental model), then the collective topological knowledge of all the nodes in your network constitute the database, and the problem reduces to remote information queries.
If you have communicated with your identity, then the information of your identity's location is cached by those nodes which conveyed the information.
I could go on for hours about it, but the basic is: it's possible, it's mathematically feasible, and it will work more or less as described.
Assuming the addresses remained relatively constant, you could assemble something after doing actual traces, yes. But precisely how constant do you figure the info for anyone concerned about traces would be?
The part I'm referring to as impossible is the idea of a database that automatically has a set list of information, rather than a constructed database like this.