Similarity between contagion and trust

[vandenheuvel]

Jointly supervised by Johan Dubbeldam and Johan Pouwelse.

Blockchains and contagion on complex networks One of the important problems which blockchain technologies attempt to solve is the problem of trust [1]. When you work for someone else, how can you be sure that you will be paid? That is, after you have worked yourself, can you trust that someone else will work for you, or give you something of value? In everyday life this problem is solved by traditional money. It is hard to duplicate, represents work performed and is controlled by a central bank.

A new blockchain model called TrustChain, developed at Delft University of Technology, attempts to solve the same problem in an entirely different, decentralized way. Transactions encoded in a public ledger give an account of all the transactions that were carried out, such that it is impossible to temper with. But unlike established blockchain models, TrustChain allows users to create and maintain their own chain, allowing the system to scale to billions of users.

This new model however is vulnerable to an attack, which requires an assessment of the trustworthiness of users. In order to aid this assessment, users exchange information about the trustworthiness of other people. The idea of information transfer on networks is not new and has been investigated in a number of papers; see e.g. [2, 3]. Moreover, it has also been shown that spreading of information is equivalent to the spreading of an epidemic on a network.

In this project we propose to investigate the Adaptive Information Diffusion (AID) model, which is equivalent to a SIS model for the spreading of epidemics. The information possession state of node in a network with nodes is specified by a Bernoulli random variable , where means that the node possesses information and when no information is present at node . At time a node possesses information with probability . We assume following [2] that only nodes that possess information can release information.

The probability for a node to possess information, that is , obeys

However, the network topology itself is also changing with time. We can take different dynamics here. One possibility would be that suggested in [2]. A link will be established between the 2 nodes and only when one of the nodes has information and the other does not; this happens with a rate . A link will be broken with rate We then obtain an equation for of the form

The effect of the dynamics can now for example be explored for different initial network topologies .

Research questions Research questions for this project are

• How well does this description apply to trust information propagation in TrustChain?
• Can the model be modified for better agreement with the TrustChain model?
• What about delays in the model? It is known that time is important in blockchain systems. How would finite spreading time effect the trustworthiness of the information? See also [4].

References

1. Van den Heuvel, B. \& Dai, Y, 2018, "Trust in Distributed Systems". Available at the repository.
2. Trajanovski, S., D. Guo and P. Van Mieghem, 2015, "From epidemics to information propagation: Striking differences in structurally similar adaptive network models", Physical Review E (R), Vol. 92, 030801(R).
3. T. Gross, C. J. D. D'Lima, and B. Blasius, 2006, Phys. Rev.Lett.96, 208701
4. R. Pass, L. Seeman, A. Shelat, 2016, Analysis of the Blockchain protocol in Asynchronous networks

This assignment as a pdf.

[qstokkink]

This issue seems to capture an individual assignment that has now been either abandoned or completed and I will close it.

[vandenheuvel]

@qstokkink this was a potential master thesis project that I was asked to document.

[qstokkink]

@vandenheuvel thanks for letting me know. If this is still an open project, I'll keep this open.