TL;DR
This paper introduces a novel covert channel that leverages Google Analytics cookies within HTTP traffic. This method is designed to be highly stealthy and offers a surprisingly high bandwidth, making it a significant threat to network security by hiding communication within seemingly innocuous web activity.
What Problem Does This Paper Address?
The paper tackles the challenge of hiding not just the content of a communication, but the very existence of that communication. While encryption secures messages, covert channels aim to conceal the fact that any data is being exchanged at all. Traditional covert channels often suffer from low bandwidth or are unidirectional. This work addresses the need for a covert channel that is bidirectional, difficult to detect, and can operate with a reasonable data rate, especially in the context of the modern, ubiquitous internet.
Methods
The core of this research is the exploitation of Google Analytics cookies, a common feature on over half of the world's websites. The authors propose using specific fields within these cookies (like utma, utmb, utmc, and utmz) to encode data. The proposed channel is bidirectional and half-duplex. Data is primarily sent from the client to the server by embedding encoded messages within the Cookie header of HTTP requests. For server-to-client communication, the server can modify the ID values in the cookies to signal a new message. The encoding scheme maps ASCII characters (32-127) to two-digit values, with a portion of the cookie dedicated to message length.
Datasets
Not discussed in the paper.
Key Insights and Ideas
- Ubiquity as a Shield: The sheer prevalence of Google Analytics makes it an ideal carrier. The massive volume of legitimate cookie traffic provides excellent cover for covert messages, making them incredibly difficult to distinguish.
- Application Layer Stealth: By operating at the application layer (HTTP cookies), the covert channel avoids detection by lower-level network monitoring tools that might focus on packet headers or timing.
- Bidirectional Potential: Unlike many previous covert channels, this method allows for communication in both directions, significantly increasing its utility for data exfiltration or command-and-control.
- High Bandwidth Potential: The use of multiple cookies and their extensive fields allows for a surprisingly high data rate, challenging the common assumption that covert channels are inherently slow.
- "Standalone" Sniffer Mode: The authors envision a mode where a "sniffer" can intercept and decode messages without the direct involvement of the Google Analytics server, further complicating detection.
Key Results
The paper reports a data rate of approximately 67 digits per HTTP request/response, translating to about 33 ASCII characters per request. Considering the full HTTP transaction, this yields an average data rate of 7.33 characters per TCP packet. For a full web transaction, they estimate a throughput of 194 characters per second in each direction. A proof-of-concept implementation in Python demonstrated the feasibility of this covert channel.
Limitations and Future Work
The authors acknowledge several limitations:
- Client-Initiated Communication: The channel is primarily client-initiated, meaning data cannot be sent to a client unless that client first makes a request.
- Party Identification: The parties involved in the communication are still identifiable, which doesn't solve the "prisoner's dilemma" in covert communication.
- Standalone Mode Limitations: The "standalone sniffer" mode, while stealthy, is unidirectional and requires the sniffer to be on the same local network.
Future work suggestions include:
- Reducing Latency: Utilizing HTTP's
keep-alivefeature to reduce the overhead of establishing new TCP connections for each message. - Enhanced Encoding: Exploring different encoding schemes to potentially increase throughput or binary file transfer capabilities.
- Leveraging Unused Fields: Investigating the
utmzandutmvcookies, which were not fully utilized in this study, to further boost bandwidth. - Improving Obfuscation: Developing more sophisticated methods to disguise the covert traffic, potentially by manipulating non-digit characters in
utmzor using theutmvcookie.
Summary updated