A dynamic state sharding blockchain architecture for scalable and secure crowdsourcing systems

Currently, the crowdsourcing system has serious problems such as single point of failure of the server, leakage of user privacy, unfair arbitration, etc. By storing the interactions between workers, requesters, and crowd sourcing platforms in the form of transactions on the blockchain, these problems can be effectively addressed. However, the improvement in total computing power on the blockchain is difficult to provide positive feedback to the efficiency of transaction confirmation, thereby limiting the performance of crowdsourcing systems. On the other hand, the increasing amount of data in blockchain further increases the difficulty of nodes participating in consensus, affecting the security of crowdsourcing systems. To address the above problems, in this paper we design a blockchain architecture based on dynamic state sharding, called DSSBD. Firstly, we solve the problems caused by cross sharding transactions and reconfiguration in blockchain state sharding through graph segmentation and relay transactions. Then, we model the optimal block generation problem as a Markov decision process. By utilizing deep reinforcement learning, we can dynamically adjust the number of shards, block spacing, and block size. This approach helps improve both the throughput of the blockchain and the proportion of non-malicious nodes. Security analysis has proven that the proposed DSSBD can effectively resist attacks such as transaction atomic attacks, double spending attacks, sybil attacks, replay attacks, etc. The experimental results show that the crowdsourcing system with the proposed DSSBD has better performance in throughput, latency, balancing, cross-shard transaction proportion, and node reconfiguration proportion, etc., while ensuring security.

Automated summary

In this paper, a blockchain architecture based on dynamic state sharding (DSSBD) is proposed to solve the problems caused by cross-shard transactions and reconfiguration. By utilizing deep reinforcement learning, the number of shards, block spacing, and block size can be dynamically adjusted to improve the performance of the blockchain. The experimental results show that the crowdsourcing system with DSSBD has better performance in terms of throughput, latency, balancing, cross-shard transaction proportion, and node reconfiguration proportion, while ensuring security.