Forensic Analysis in Access Control: a Case-Study of a Cloud Application
Bibliographic record
Abstract
We discuss a case-study we have conducted on forensic analysis in access control. The case-study is an application in the Amazon Web Services (AWS) cloud provider. Forensic analysis is the investigation and analysis of evidence of possible wrongdoing. Access control is used to regulate accesses to computing resources. Both forensic analysis and access control are recognized as important aspects of the security of a system. We first argue that posing the forensic analysis problem in the context of access control is meaningful and useful towards the security of a system. We then summarize results on the computational hardness of the forensic analysis problem for two access control schemes from the research literature. We point out that these results suggest that meaningful logging information can render forensic analysis tractable, even efficient. We then instantiate the forensic analysis in access control problem in the context of a cloud application. A cloud application is a software service that can be accessed over the Internet and uses computing resources provided by a cloud provider. A cloud provider provides computing tools and services that can be administered over the Internet. The cloud provider we have adopted is AWS, and the application is ``Hello Retail'', an image-sourcing application for online retailers. In addressing forensic analysis in this context, our particular focus is the manner in which logging information can be leveraged. We ask two kinds of questions: (i) is particular logging information from AWS necessary to answer forensics analysis questions of interest, and, (ii) is particular logging information sufficient? We observe that from the standpoint of (i), default AWS logs have considerable redundancy. We propose an algorithm to prune logs for efficient forensic analysis. From the standpoint of (ii), we observe that it is not possible to definitively answer "yes" or "no" to forensic analysis questions of interest given only the information AWS permits us to log. We identify additional logging information that, if available, would be sufficient. Together, (i) and (ii) provide us with "goal-directed logging". We conclude by reiterating the benefits of forensic analysis in access control, and with suggestions for goal-directed logging in cloud systems.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".