Extractocol: Automatic Extraction of Application-level Protocol Behaviors for Android Applications

摘要

Android app is an important class of today's Internet applications that generate roughly 40-50% of mobile Web and app traffic. More than 1.4 million Android apps are offered through Google's open market, and tens of thousands of new apps are added every month. However, very little information is known about their application protocol behaviors because they predominantly use proprietary protocols on top of HTTP. The problem is further exacerbated by the popular use of common data representation, such as JSON and XML, due to the popularity of REST-ful web services. As a result, application protocols appear similar to each other, and analyzing them requires an in-depth characterization of each individual application. Despite this, the state of the art remains that even the problem of finger-printing the traffic they generate is extremely challenging, let alone a full protocol analysis.