DBFT Leader Selection Hashing to Reduce Cumulative Delay
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
DBFT distributed networks experience significant delays in block generation due to multiple continuous node failures, as the existing leader selection mechanism leads to cumulative delay times when multiple nodes fail consecutively.
Innovation Solution
A method is introduced to alter the leader selection process by directly incrementing the last leader index when the view index is zero and calculating a new leader index using a merged Hash value of block height and view index through the blake2b algorithm when the view index is not zero, ensuring that block generation delays are minimized.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the traditional leader selection formula p=(h+v) % n is used in DBFT, then the leader selection follows a predictable sequential pattern, but when multiple continuous nodes fail, the block generation delay accumulates to m*t where m is the number of failed nodes
Solution Approach 1:
The patent changes the leader selection parameter calculation by introducing a hash-based random selection mechanism instead of the traditional sequential formula. When a leader failure is detected, the system calculates a new leader index using hash(h, v, failed_leader_index) % n, where h is block height, v is view index, and n is total nodes. This parameter change ensures that leader selection is no longer predictable and sequential, thereby preventing cumulative delays when multiple continuous nodes fail.
2Productivity
If the view index is incremented sequentially (v++) upon leader failure, then the system attempts to elect a new leader, but the sequential increment causes repeated failures when multiple continuous nodes are down
Solution Approach 1:
The patent implements preliminary action by pre-calculating potential leader candidates using hash-based random selection before actual leader failure occurs. When a failure is detected, the system already has a predetermined alternative leader index ready through the hash calculation, eliminating the need for sequential view index increments and repeated election attempts. This preliminary preparation of alternative leaders directly reduces the cumulative delay that would otherwise occur with sequential v++ increments.
3Device complexity
If the DBFT system uses sequential leader election with view index increment, then the system maintains a simple election mechanism, but the block generation delay becomes m*t for m continuous node failures
Solution Approach 1:
The patent replaces the mechanical sequential increment mechanism (v++) with a hash-based random selection mechanism. Instead of mechanically incrementing the view index and sequentially trying the next leader, the system uses cryptographic hash functions to randomly select alternative leaders. This substitution maintains acceptable system complexity while dramatically reducing block generation delay from m*t to approximately t, as the hash-based selection can quickly identify a valid alternative leader without sequential trial and error.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively reduces the cumulative delay time caused by multiple continuous node failures, thereby accelerating block generation in DBFT distributed networks.
Implementation Method 1
Hashing the merged binary stream bhv and obtaining a merged binary stream Hash value
Data Source
AI summary
A method for weakening delays resulted from failures of multiple continuous nodes in a DBFT distributed network is provided. In step S1, whether a view index is equal to zero is determined, if yes, a new Leader index is obtained according to a first rule and the method processes to step S3, otherwise processes to step S2. In step S2, the new Leader index is obtained according to a second rule, and whether a proposal proposed by a node corresponding to the new Leader index is passed is determined. If yes, the method processes to step S3, otherwise, step S2 is implemented repeatedly until the proposal proposed by the node corresponding to the new Leader index is passed. In step S3, the node corresponding to the new Leader index is taken as a new Leader. Then the delays caused by multiple continuous nodes in the DBFT distributed network are weaken.