Some blockchain systems have used Layer2 to improve the performance on chain. However, most of the existing Layer2 schemes assume that the off-chain nodes are infinite, but when the resources off chain are limited, the system needs to re-measure the off-chain load and dynamically allocate the number of transactions on and off chain. While off-chain resources are limited, how to balance the load on and off chain to maximize the use of Layer2 is an urgent problem. Therefore, this paper regards the on and off-chain nodes as two parties in Stackelberg Game, and dynamic Game obtains a reasonable division of the number of transactions. First, multiple parameters are set to establish the model, and the influence of several factors on the Game model is analyzed. Then the parameters are constantly refined to analyze the best decision of the Game more accurately. The experiment proves the effectiveness of the Game, the throughput of the baseline after the Game is about 77% higher, the latency is about 52% lower. The experiments also give the optimal on and off-chain decisions under different parameters.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

On and Off-Chain Load Balancing Model Based on Stackelberg Game

  • Xuyang Liu,
  • Lanju Kong,
  • Yuehan Su,
  • Wei Guo

摘要

Some blockchain systems have used Layer2 to improve the performance on chain. However, most of the existing Layer2 schemes assume that the off-chain nodes are infinite, but when the resources off chain are limited, the system needs to re-measure the off-chain load and dynamically allocate the number of transactions on and off chain. While off-chain resources are limited, how to balance the load on and off chain to maximize the use of Layer2 is an urgent problem. Therefore, this paper regards the on and off-chain nodes as two parties in Stackelberg Game, and dynamic Game obtains a reasonable division of the number of transactions. First, multiple parameters are set to establish the model, and the influence of several factors on the Game model is analyzed. Then the parameters are constantly refined to analyze the best decision of the Game more accurately. The experiment proves the effectiveness of the Game, the throughput of the baseline after the Game is about 77% higher, the latency is about 52% lower. The experiments also give the optimal on and off-chain decisions under different parameters.