In a payment channel network, Alice can transfer money to Bob by using intermediate nodes'

Contexts in source publication

Context 1

... Bob A payment channel network is a collection of bidirectional payment channels (Fig. 2). If Alice wants to send three tokens to Bob, she first finds a path to Bob that can support three tokens of payment. Intermediate nodes on the path (Charlie) will relay payments to their destination. Hence in Fig. 2, two transactions occur: Alice to Charlie, and Charlie to Bob. To incentivize Charlie to participate, he receives a ...

Context 2

... Bob A payment channel network is a collection of bidirectional payment channels (Fig. 2). If Alice wants to send three tokens to Bob, she first finds a path to Bob that can support three tokens of payment. Intermediate nodes on the path (Charlie) will relay payments to their destination. Hence in Fig. 2, two transactions occur: Alice to Charlie, and Charlie to Bob. To incentivize Charlie to participate, he receives a routing fee. To prevent him from stealing funds, a cryptographic hash lock ensures that all intermediate transactions are only valid after a transaction recipient knows a private key generated by Alice [18]. 1 Once Alice ...

Context 3

... a fluid approximation model for Spider's dynamics, we can show that the rates computed by Spider are an optimal solution to the routing problem in Equations (1)-(5) for parallel networks (such as Fig. 20 in App. B). In the fluid model, we let x p (t) denote the rate of flow on a path p at time t; for a channel (u, v), f u,v (t) denotes the fraction of packets that are marked at router u as a result of excessive queuing. The dynamics of the flow rates x p (t) and marking fractions f u,v (t) can be specified using differential equations ...

Context 4

... e n } that are connected via k parallel payment channels (r 1 , r 1 ), . . . , (r k , r k ) as shown in Figure 20. The end-hosts from each set have demands to end-hosts on the other set. ...

Context 5

... run five circulation traffic matrices for 1010s on our three topologies with all channels having exactly the tokens denoted by the channel size. Fig. 23 shows that across all topologies, Spider outperforms the state-of-the-art schemes on success ratio. Spider is able to successfully route more than 95% of the transactions with less than 25% of the capacity required by LND. Further Fig. 24 shows that Spider completes nearly 50% more of the largest 12.5% of the transactions attempted in ...

Context 6

... is able to successfully route more than 95% of the transactions with less than 25% of the capacity required by LND. Further Fig. 24 shows that Spider completes nearly 50% more of the largest 12.5% of the transactions attempted in the PCN across all three topologies. Even the waterfilling heuristic outperforms LND by 15-20% depending on the topology. ...

Context 7

... §7.3, we show that Spider outperforms state-of-the art schemes on the success ratio achieved for a given channel capacity. Here, we break down the success volume by flows (sender-receiver pairs) to understand the fairness of the scheme to different pairs of nodes transacting on the PCN. Fig. 25 shows a CDF of the absolute throughput in e/s achieved by different protocols on a single circulation demand matrix when each sender sends an average of 30 tx/s. The mean channel sizes for the synthetic topologies and the real topologies with LCSD channel sizes are 4000e and 16880e respectively. We run each protocol for 1010s and ...

Context 8

... the achieved throughput is closer to the maximum when there is a smaller component of DAG in the demand matrix. This suggests again that the DAG affect PCN balances in a way that also prevents the circulation from going through. We investigate what could have caused this and how pro-active on-chain rebalancing could alleviate this in §7.4. Fig. 27 shows the success ratio and normalized throughput achieved by different schemes when rebalancing is enabled for the 20% DAG traffic demand from Fig. 26. Spider is able to achieve over 95% success ratio and 90% normalized throughput even when its routers balance only every 10,000 e while LND is never able to sustain more than 75% ...

Context 9

... affect PCN balances in a way that also prevents the circulation from going through. We investigate what could have caused this and how pro-active on-chain rebalancing could alleviate this in §7.4. Fig. 27 shows the success ratio and normalized throughput achieved by different schemes when rebalancing is enabled for the 20% DAG traffic demand from Fig. 26. Spider is able to achieve over 95% success ratio and 90% normalized throughput even when its routers balance only every 10,000 e while LND is never able to sustain more than 75% success ratio even when rebalancing for every 10e routed. This implies that Spider makes PCNs more economically viable for both routers locking up funds in ...

Context 10

... payment channel network is a collection of bidirectional payment channels (Fig. 2). If Alice wants to send three tokens to Bob, she first finds a path to Bob that can support three tokens of payment. Intermediate nodes on the path (Charlie) will relay payments to their destination. Hence in Fig. 2, two transactions occur: Alice to Charlie, and Charlie to Bob. To incentivize Charlie to participate, he receives a ...

Context 11

... payment channel network is a collection of bidirectional payment channels (Fig. 2). If Alice wants to send three tokens to Bob, she first finds a path to Bob that can support three tokens of payment. Intermediate nodes on the path (Charlie) will relay payments to their destination. Hence in Fig. 2, two transactions occur: Alice to Charlie, and Charlie to Bob. To incentivize Charlie to participate, he receives a routing fee. To prevent him from stealing funds, a cryptographic hash lock ensures that all intermediate transactions are only valid after a transaction recipient knows a private key generated by Alice [5]. Once Alice is ...