This during taint propagation as TaintDroid storesThis during taint propagation as TaintDroid stores

This research paper aims to resolve the problem regarding the usage of private data in a mobile OS. The paper acknowledges that the control of access to private information is provided to a user but it is not revealed how the accessed private information is used or where it is sent. The paper solves this problem by devising TaintDroid which is a taint tracking system and also has the capability to analyze various data sources concurrently. The paper achieves this by providing the mobile app user with a feedback which includes the taints to data along with application information & destination of tainted data. The novelty introduced by the paper which other taint trackers do not cover is the system-wide tracking.The paper contributes by combining the four existing techniques(variable, method, message and file level) using android architecture to determine the balance between performance and accuracy. The tainting is stored with context and not just as random labels. The proposed solution of paper minimizes storage overhead by storing one taint tag per array in TaintDroid. By implementing message level taint propagation TaintDroid prevents taint tag removal by applications.The paper does not address control flow tracking i.e. it does not provide the solution to data leaks via implicit flows. The paper revels the risk of generating false positive results during taint propagation as TaintDroid stores only one taint tag per array. The overhead regarding the address book read & write has not been discussed in detail in the paper. TaintDroid does not include tracking of tainted information received in the phone in the form of network reply. Another source which generates false positives is when information has configuration identifiers and this could potentially change results.The follow-up work could include the focus to minimize the chances of false positives which occur due to storage of 1 taint tag per array and due to message level propagation. Word level taint tags can be explored for the precise propagation of taint tags. However, this might affect performance so follow up work could include determining the trade-off between these aspects.