Towards Fuzz Testing a Procedurally-Generated Video Game

Erik Fredericks

Download Paper | Permalink

Conference: 2024 ASEE North Central Section Conference
Location: Kalamazoo, Michigan
Publication Date: March 22, 2024
Start Date: March 22, 2024
End Date: March 23, 2024
Page Count: 10
DOI: 10.18260/1-2--45643
Permanent URL: https://peer.asee.org/45643
Download Count: 15

Request a correction

Paper Authors

biography

Erik Fredericks Grand Valley State University

visit author page

Erik Fredericks is an Assistant Professor in the School of Computing at Grand Valley State University. His research focuses on exploring how uncertainty can impact self-adaptive and safety-critical systems at different levels of abstraction and how it can be mitigated by using search-based software engineering techniques. Recently, he has been investigating how generative art can be automatically created via evolutionary computation.

visit author page

Download Paper | Permalink

Abstract

Fuzz testing presents opportunities for discovering bugs in software projects that are unanticipated by developers as large amounts of either random or targeted inputs are applied to the system under test. Moreover, exploratory techniques such as search-based fuzz testing can discover new and interesting combinations of input data that can further lead to bug discovery. Video games are a subset of software projects that involve the additional overhead of audio/visual cues for gameplay, state management, and rigorous timing constraints. Procedural content generation (PCG) can be used to support development by incorporating unique game content (e.g., items, storylines, environments, etc.) via algorithms. As such, verification of PCG techniques is necessary to ensure that the generated content is valid for the situations in which they are deployed, given that such content can lead to emergent gameplay (i.e., unanticipated interactions that result in new features) or user dissatisfaction (e.g., the "same" type of rock is generated multiple times in a small area). We present our work-in-progress efforts and proposed run-time software testing methodology for developing an experimental testbed for fuzzing procedural generation in video games.

This project was created as part of the NSF RISE program for first-generation students to participate in an active research program. Delve the Dungeon is our prototype framework for exploring how software engineering can enhance assurance that PCG techniques are executing as expected. Specifically, this framework provides a roguelike-style video game environment that comprises procedurally generated dungeons and text, with common features of this particular game domain including turn-based gameplay, bump-to attack, and different forms of monsters that attack the player. Additionally, we have developed a proof-of-concept requirements specification to support our software engineering activities, where the next phase of development will monitor those requirements at run time, use the requirements as a basis for creating test cases and generating fuzzed test data, and then incorporate the results of run-time requirements monitoring and test case execution to the application as part of a feedback loop to continuously improve its behavior.

Citation
Format

Fredericks, E. (2024, March), Towards Fuzz Testing a Procedurally-Generated Video Game Paper presented at 2024 ASEE North Central Section Conference, Kalamazoo, Michigan. 10.18260/1-2--45643

TY  - CPAPER
AB  - Fuzz testing presents opportunities for discovering bugs in software projects that are unanticipated by developers as large amounts of either random or targeted inputs are applied to the system under test.  Moreover, exploratory techniques such as search-based fuzz testing can discover new and interesting combinations of input data that can further lead to bug discovery.    Video games are a subset of software projects that involve the additional overhead of audio/visual cues for gameplay, state management, and rigorous timing constraints.  Procedural content generation (PCG) can be used to support development by incorporating unique game content (e.g., items, storylines, environments, etc.) via algorithms.  As such, verification of PCG techniques is necessary to ensure that the generated content is valid for the situations in which they are deployed, given that such content can lead to emergent gameplay (i.e., unanticipated interactions that result in new features) or user dissatisfaction (e.g., the &quot;same&quot; type of rock is generated multiple times in a small area).   We present our work-in-progress efforts and proposed run-time software testing methodology for developing an experimental testbed for fuzzing procedural generation in video games.

This project was created as part of the NSF RISE program for first-generation students to participate in an active research program.  Delve the Dungeon is our prototype framework for exploring how software engineering can enhance assurance that PCG techniques are executing as expected.  Specifically, this framework provides a roguelike-style video game environment that comprises procedurally generated dungeons and text, with common features of this particular game domain including turn-based gameplay, bump-to attack, and different forms of monsters that attack the player.  Additionally, we have developed a proof-of-concept requirements specification to support our software engineering activities, where the next phase of development will monitor those requirements at run time, use the requirements as a basis for creating test cases and generating fuzzed test data, and then incorporate the results of run-time requirements monitoring and test case execution to the application as part of a feedback loop to continuously improve its behavior.
AU  - Erik Fredericks
CY  - Kalamazoo, Michigan
DA  - 2024/03/22
PB  - ASEE Conferences
TI  - Towards Fuzz Testing a Procedurally-Generated Video Game
UR  - https://peer.asee.org/45643
DO  - 10.18260/1-2--45643
ER  -