Not logged in.
Quick Search - Contribution
Contribution Details
| Type | Master's Thesis |
| Scope | Discipline-based scholarship |
| Title | Decentralized Reputation in Blockchain Oracle |
| Organization Unit | |
| Authors |
|
| Supervisors |
|
| Language |
|
| Institution | University of Zurich |
| Faculty | Faculty of Business, Economics and Informatics |
| Date | 2023 |
| Abstract Text | This thesis presents the design of a decentralized reputation system situated within a data on-chaining solution built atop Substrate. In particular, it is a response to the quest for an ideal design of a reputation system for a blockchain oracle. We assume the context of a concrete instantiation of such an oracle, Acurast, and situate our reputation system therein. By means of extensively consulting the literature, frequently encountered threats to and vulnerabilities of present day reputation systems formed the basis for a requirements elicitation, along with technical demands stemming from the technical context in which the proposed solution is situated. Based on these requirements and building on well-established principles of the reputation system literature, we present a design and show how it satisfactorily addresses the previously defined requirements, related to security properties as well as performance measures. Our proposed reputation engine makes use of probability distributions to model reputation scores as the posteriori probabilities of binary events. We thus achieve a design that is highly performant, carries low memory overhead and is inherently immune to a number of well-known threats commonly mentioned in the literature. We thus put forward arguments for having come up with a quasi-ideal design, despite acknowledging that no IT system can be deemed perfect. Further, we show by means of agent-based modeling how a highly relevant expected failure rate within Acurast, can be reduced by almost half. Additionally, we show the reputation system’s simulated effect on the matching algorithm of Acurast. |
| PDF File |
Download
|
| Export |
BibTeX
|