Is this project an undergraduate, graduate, or faculty project?
Undergraduate
individual
What campus are you from?
Daytona Beach
Authors' Class Standing
Austen Pallen, Senior
Lead Presenter's Name
Austen Pallen
Faculty Mentor Name
Dr. Bryan Watson
Abstract
New technologies rely on multi-agent systems (MAS’s) and their ability to achieve consensus. Current distributed consensus approaches, however, have narrow applicability and are only resilient to a small subset of faults. Biologically Inspired Design may provide the inspiration needed to develop more applicable consensus algorithms. Our hypothesis is that if the biological behavior of synchronous turtle hatching is evaluated, then a more resilient and novel consensus algorithm can be developed, because current turtle hatching requires resilient consensus for species survival. To test this, an Agent-Based, ANYLOGIC model was developed based on the turtle behavior and tested against 1, 5, 10, 15, and 20 faulted agent(s) across four different environments. The time taken for 66% of the agents to accurately reach consensus about environmental conditions was recorded. There were 50 runs per faulted agent per environment totaling 1,200 runs. The agreement time average for the tests that consistently reached the consensus limit had coefficients of variance below 6%, showing resilience to faulted agents and proving that the proposed distributed consensus algorithm was resilient to faulted agents (even up to 20% of the population). Additionally, the results provide insight into the type of scenario the algorithm can be applied to (minimum viable parameter rate of change requirement).
Did this research project receive funding support from the Office of Undergraduate Research.
Yes, SURF
A New Biologically Inspired Consensus Algorithm Protocol for Multi-Agent Systems based on Turtle Hatching
New technologies rely on multi-agent systems (MAS’s) and their ability to achieve consensus. Current distributed consensus approaches, however, have narrow applicability and are only resilient to a small subset of faults. Biologically Inspired Design may provide the inspiration needed to develop more applicable consensus algorithms. Our hypothesis is that if the biological behavior of synchronous turtle hatching is evaluated, then a more resilient and novel consensus algorithm can be developed, because current turtle hatching requires resilient consensus for species survival. To test this, an Agent-Based, ANYLOGIC model was developed based on the turtle behavior and tested against 1, 5, 10, 15, and 20 faulted agent(s) across four different environments. The time taken for 66% of the agents to accurately reach consensus about environmental conditions was recorded. There were 50 runs per faulted agent per environment totaling 1,200 runs. The agreement time average for the tests that consistently reached the consensus limit had coefficients of variance below 6%, showing resilience to faulted agents and proving that the proposed distributed consensus algorithm was resilient to faulted agents (even up to 20% of the population). Additionally, the results provide insight into the type of scenario the algorithm can be applied to (minimum viable parameter rate of change requirement).