NoMoAds: Effective and Efficient Cross-App Mobile Ad-Blocking

Abstract

Although advertising is a popular strategy for mobile app monetization, it is often desirable to block ads in order to improve usability, performance, privacy, and security. In this paper, we propose NoMoAds to block ads served by any app on a mobile device. NoMoAds leverages the network interface as a universal vantage point: it can intercept, inspect, and block outgoing packets from all apps on a mobile device. NoMoAds extracts features from packet headers and/or payload to train machine learning classifiers for detecting ad requests. To evaluate NoMoAds, we collect and label a new dataset using both EasyList and manually created rules. We show that NoMoAds is effective: it achieves an F-score of up to 97.8% and performs well when deployed in the wild. Furthermore, NoMoAds is able to detect mobile ads that are missed by EasyList (more than one-third of ads in our dataset). We also show that NoMoAds is efficient: it performs ad classification on a per-packet basis in real-time. To the best of our knowledge, NoMoAds is the first mobile ad-blocker to effectively and efficiently block ads served across all apps using a machine learning approach.

AdGraph: A Graph-Based Approach to Ad and Tracker Blocking

“…Ad and tracker blocking is a well studied topic (e.g. [36], [37], [45], [46], [49], [58], [64], [65]). However, existing work is insufficient to form a comprehensive and robust blocking solution.…”

“…More significantly, much related work proposes resource blocking strategies, but without an evaluation of how their blocking strategy would affect the usability of the web. To name some examples, [36], [64], [58], [49], [46], [45], and [37], all propose strategies for automatically blocking web resources in pages, without determining whether that blocking would harm or break the user-serving goals of websites ( [65] is an laudable exception, presenting an indirect measure of site breakage by way of how often users disabled their tool when browsing). Work that presents how much bad website behavior an approach avoids, without also presenting how much beneficial behavior the approach breaks, is ignoring one half of the ledger, making it difficult to evaluate each work as a practical, deployable solution.…”

The Representativeness of Automated Web Crawls as a Surrogate for Human Browsing

“…Of course, crawls are not the only way to perform Web measurement studies. Other works have built datasets based on proxy interceptions [22,24,45], instrumented networks and VPNs [24,45,53]. Papadopoulos et al use a dataset collected over 1 year from 850 participants and explicitly state they are not affected by distortions that exist in crawled data [45].…”

WebGraph: Capturing Advertising and Tracking Information Flows for Robust Blocking

“…Prior ML-based approaches mainly detect ads and trackers at the network and JavaScript layers of the web stack. Specifically, these approaches detect ads and trackers by featurizing network requests [10,11,12] or JavaScript code [13,14,15].…”

“…The research community is actively developing machine learning (ML) approaches to automate the detection of advertising and tracking and make filter lists more comprehensive. The first generation of ML-based blocking approaches analyze network requests [10,11,12] or JavaScript code [13,14,15] to learn distinctive behaviors of advertising and tracking. However, these ML-based blocking approaches are highly susceptible to adversarial evasion techniques that are already found in the wild, including URL obfuscation [16] and code obfuscation [17].…”