In the policies of AdMob, it is stated that -
Publishers may not click their own ads or use any means to inflate impressions and/or clicks artificially, including manual methods. Testing your own ads by clicking on them is not allowed.
But how does AdMob detects if developer clicks his own ad? I have read about many incidents where developer's money is confisticated.
Do they track IP address of developer? Do they track some other things too in order to keep track of developer clicks?
Google (which owns AdMob) does not disclose the exact details of how click fraud is detected (because revealing all the details would make it easy to circumvent). However, they do occassionally reveal some of the general underlying principles or some of the factors involved as they did in this 2008 blog post. In terms of automated detection, the general priniciple is anomaly detection and, as stated in the blog post, the IP address is one (of many) factors that are considered in it.