An Optimization of DPoS for Swarm Intelligence
Kailei Tang,Zhiyan Dong,Tianlun Dai,Zhongxue Gan
DOI: https://doi.org/10.1109/cac53003.2021.9728381
2021-01-01
Abstract:The issues of managing swarm intelligence are essential to many multiple tasks. The scenarios are complex and dynamic, which is hard for a single agent to satisfy the needs of various tasks. As thus, a practical intelligence cooperative combat scheme, composed of multiple agents, is required to supply a effective and efficient consensus for swarm intelligence with external conditions evolving. Among this scheme, the accuracy of information sharing, transmission, and the integrity of the stored data are all critically significant. The Blockchain, a digital distributed ledger,is distributed on different nodes, and each node holds the same data, which attracts increasing attentions recently. There are many consensus algorithms which is the key part of the blockchain, such as PoW, PoS, DPoS, PoR, etc. However, no matter in terms of performance, security and stability, existing work can hardly support a battle plan oriented to swarm intelligence. To our knowledge, there is no consensus algorithm that takes into account the resources of agent in swarm intelligence collaboration. Therefore, we introduce an optimization of DPoS for swarm intelligence based on agent behavior monitoring and agent’s own resource analysis (Delegated Proof of Stake based node’s Behavior and Resource, DPoSBR). Combining the situation of malicious behaviors of the agent and the resources of the agent, we choose the more trustworthy agent as the captain. Therefore, the captain agent is more secure and the election process is fairer. Finally, the extensive simulations are conducted to evaluate the performance of DPoSBR algorithm, which has good practicability. Meanwhile, it enables more agents to participate, which is beneficial to the decentralization of the system and can promote the enthusiasm of the entire agents, and it prevents the malicious agents from doing malicious behaviors again.