Byzantine General Problem

From the above consensus that the set of verifiers can form a distributed system. Assuming that the delay of the system network is L, the following conclusions can be drawn from the previous BFT study:

1. In a partially synchronized system (ie, L<d, d is a constant, but the boundary value of d is not known), if a node is less than 1/3 evil, or does not respond, the system can reach a consensus. For example, the PBFT algorithm can reach a consensus when less than 1/3 of the nodes are not responding.

2. The asynchronous model (that is, L is infinite) has no consensus method.

Assume that an attacker uses a network attack (DDoS) to make at least 1/3 of nodes unable to respond, so that a distributed system composed of verifiers cannot reach a consensus and the transaction can never be verified.

Based on the above assumption, Ambr sets a boundary value (1 hour or 1 day). The system will determine that the distributed system composed of the set of verifiers cannot reach consensus. Ambr's response strategy is as follows:

1. Ambr will reassemble the verifier collection, or replace some unresponsive nodes, or combine a new one.

2. All nodes in the system will receive instructions to inform the verifier that the set cannot reach consensus. At this time, the system sets a delay time, assume that this time is T.

3. The node starts the timer after receiving a new verification module. When the time t > T is received, the node can determine that the verification module is valid.

4. The system investigates the original set of verifiers. If the survey results indicate no cyber-attack, the verifier is not responding to its own cause, and the verifier will be punished accordingly. If the investigation results in an attacker launching a cyberattack, all of the attacker’s deposits will be forfeited.