In intelligent transportation systems

vehicles communicate with nearby roadside units to share real-time information such as vehicle ID

location

speed

and status. This data is transmitted through open channels

raising concerns about trust and privacy. To address these challenges

this paper introduces a distributed

immutable

and secure blockchainsharding network

which is deployed in the vehicle-road collaboration scenario. A collaborative edge-cloud architecture is proposed

consisting of the edge device layer

the roadside unit layer

and the in-vehicle terminal layer. The edge devices form the main chain

roadside units act as shards

and vehicles accessing the system become accounts. Although sharding enhances the scalability of the system

the imbalance of workload between shards can lead to congestion in some shards. Moreover

the limited computational capacity of roadside units may result in resource wastage if resources are not dynamically allocated according to load. This paper proposes a load balancing mechanism based on vehicle-road collaboration to optimize the allocation of computational resources in sharding. The load balancing and computational power allocation metrics of the system are established. Cooperative multi-agent reinforcement learning is used to optimize account distribution and computational resource allocation

achieving load balancing and efficient resource utilization. Experimental results show that the proposed approach significantly reduces the imbalance of load between shards and reduces resource consumption

while ensuring the security of system. Compared to traditional sharding strategies

the throughput of the system increases by over 15%

demonstrating superior performance.