Travis T. Schluessler

Is a bot at the controls?: Detecting input data attacks

The use of programmatically generated input data in place of human-generated input data poses problems for many computer applications in use today. Mouse clicks and keyboard strokes can automatically be generated to cheat in online games, or to perpetrate click fraud. The ability to discern whether input data was computationally generated instead of created by a human input device is therefore of paramount importance to these types of applications. This paper describes a method for detecting input data that was computationally modified or fabricated. This includes detecting data that was not directly generated by a physical human input device such as a keyboard or mouse. A prototype of this system was built on existing hardware and was shown to be effective at detecting attacks on a real application. This detection method is capable of addressing the majority of input-based attacks currently in use. When used in conjunction with a trusted peripheral, it offers a robust mechanism for ensuring a computer is not at the controls.

Stealth measurements for cheat detection in on-line games

As a result of physically owning the client machine, cheaters in network games currently have the upper-hand when it comes to avoiding detection by anti-cheat software. To address this problem and turn the tables on cheaters, this paper examines an approach for cheat detection based on the use of stealth measurements via tamper-resistant hardware. To support this approach, we examine a range of cheat methods and a number of measurements that such hardware could perform to detect them.

The case for network witnesses

Network abuse is prevalent in todaypsilas Internet. To combat abuse, this paper examines a general approach for constructing network protocols based on the use of ldquonetwork witnessesrdquo: tamper-resistant, trusted third parties that reside at network protocol end-points. By providing authentic measurements of network use and by ensuring the correct usage of network protocols, we show how network witnesses can enable fundamentally new protocol designs that can protect networks against malicious use.

Limiting CPU power consumption for efficient computation of 3D workloads

Rendering 3D workloads using the least power possible is an increasingly important quality of computing platforms. Current platforms do not achieve this goal because they power the Central Processing Units (CPUs) at frequencies above the minimum required for these workloads to operate without performance loss. Higher than necessary frequencies yield greater than necessary power consumption. This paper describes a method for limiting CPU frequency while running 3D workloads to reduce power consumption with minimal performance loss. Using this method on an Intel® 3rd generation CoreTM processor reduces CPU power consumption by an average of 9% with no significant performance impact.