Designing and Evaluating Rebalancing Algorithms for Payment Channel Networks

Designing and Evaluating Rebalancing Algorithms for Payment Channel Networks

van Engelshoven, Yuup (TU Delft Electrical Engineering, Mathematics and Computer Science)

Roos, Stefanie (mentor)
Epema, Dick (graduation committee)
Pawelczak, Przemek (graduation committee)

Delft University of Technology

2019-12-10

Payment Channel Networks(PCN) utilize payment channels with an established link capacity between two nodes to route transactions over multiple links to carry out transactions. Such transactions can support a blockchain due to the transactions happening off-chain, i.e., not requiring any information to be published to a ledger. PCNs can help aid in the scalability of blockchains, by moving transactions off-chain not all transactions need to be stored on the blockchain, reducing the amount of data that needs to be stored on the blockchain. Lightning is the PCN implementation that makes use of Bitcoins blockchain. As transactions occur over the network the capacity of the link may vary over time between two nodes. This change may lead to the link being only available from one side. If enough links become unavailable then processing transactions may take longer or in the worst-case scenario transactions may no longer be feasible in the network. To help avoid these short-comings in PCN strategies can be designed in path-based transaction algorithms to help keep links capable of handling transactions bidirectionally. This work presents two such algorithms, the Passive Merchant and Active Merchant. Additionally, two synthetic data-set models are proposed to help evaluate the effectiveness of the Merchant algorithms, due to a lack of data-sets in this field. In the evaluation two different topologies are examined to evaluate the impact a graph has on the success rate of transactions within a PCN. The evaluation of the The Merchant algorithms is simulation based, experiments evaluated how different algorithms effected the success rate of transactions. The simulation did indicate that the algorithms were able to help increase the success rate of transactions, up to 8%. As these algorithms are embedded in the transaction process of a payment the algorithms are a first of there kind, other solutions have been proposed for rebalancing as a separate protocol. In addition to being the first to propose transaction embedded rebalancing algorithms, no other synthetic data-set models for PCN have been proposed. The synthetic data-set models may allow this area of research to be have constant data-sets that are used to evaluate the effectiveness of path-based transactions.

Blockchain
Payment Channel Network
SpeedyMurmurs
Data-set
Synthetic Data-set

http://resolver.tudelft.nl/uuid:32bf01ba-b6c6-410e-b0e1-b15d95eb05d9

2019-12-05

Student theses

master thesis

© 2019 Yuup van Engelshoven