TU Berlin

Service-centric NetworkingWenzel, D. (2019). An Entity-Taxonomy-based Analysis of Token Networks. Master Thesis, Technische Universität Berlin

Page Content

to Navigation

Master Thesis: An Entity-Taxonomy-based Analysis of Token Networks


An Entity-Taxonomy-based Analysis of Token Networks


Over the past decade, distributed ledger technology and blockchain emerged. An innovation enabled by these technologies is tokens. Tokens are digital assets which ownership rules are decentrally enforced through a blockchain. Access and property rights can be bound to those tokens, or they can derive value from being used in novel decentralized applications. How tokens can be used is a subject of ongoing research. In this work, a taxonomy-based method for empirically analyzing how tokens are used is proposed and evaluated by creating an entity taxonomy of all entities in the ERC20 token network.

On major blockchains like Ethereum and Bitcoin, transfers occur between pseudonymous addresses and a token network can be constructed out of it. Using heuristics, some address can be clustered into entities. This work proposes to create a taxonomy of these entities, in which their behavior is characterized along dimensions like age, holding size and method of operation. The usefulness of the method is shown by creating a taxonomy of ERC20 token users. ERC20 is a standard for fungible Ethereum tokens and is implemented by the majority of exchange-traded second-layer tokens. In the taxonomy process, this work proposes a set of novel heuristics for clustering ERC20 address into entities.

As shown by example, the ERC20 entity taxonomy can highlight differences in the entities using different tokens like stable token and payment token. With the taxonomy, upper bounds of the total amount of token units used for their intended purpose is calculated for two tokens. The bounds are at 1.5% and 0.88%. The taxonomy can characterize active entities and pas- sive investors and highlight that the distribution of these groups over all token classes are not uniform. Using k-anonymity, this work shows that an adversary who can guess some of the taxonomy characteristics of an entity is more likely to link them with other information, which questions the contextual integrity of ERC20 token users. 

Supervisor: Marcel Müller, Dr. Peter Ruppel

Type:  Master Thesis

Duration: 4 months


Quick Access

Schnellnavigation zur Seite über Nummerneingabe