Blockchain-based 5G Wireless Access Network Resource Sharing Framework and Secure Resource Allocation Method

Abstract

In response to problems such as a lack of trust, low resource utilization rates, conflicts due to multiple constraints, and security risks associated with sharing 5G wireless access network resources, this study proposes an efficient, trustworthy, and secure distributed resource sharing system and optimizes the resource allocation strategy. First, it performs virtual decoupling and atomic modeling for the three core computing resources: spectrum, security, and computing power. It also designs a five-layer distributed resource-sharing framework that integrates blockchain and software-defined networks. Additionally, it proposes an improved delegated proof-of-stake consensus mechanism, as well as an asymmetric encryption transaction authentication and resource status traceability mechanism. Second, for the multi-constraint conflict issue, it designs a multi-agent deep deterministic strategy gradient secure resource allocation algorithm integrating long-term and short-term memory state prediction. The verification experiments were carried out based on the 5G-RAN public resource scheduling dataset in accordance with the 3GPP TR38.901 protocol specification. The experimental hardware was equipped with Intel Core i9-13900K processor, NVIDIA RTX 4090 graphics card, etc. The simulation platform was built on the Ubuntu 22.04 LTS system using the PyTorch 2.1.0 deep learning framework and the NS-3 3.36 simulation tool. The comparison benchmarks were mainstream centralized resource allocation schemes, blockchain, federated deep reinforcement learning schemes, and consortium chain hierarchical cross-slice schemes. The experimental results showed that the resource utilization rate of this framework reached 89.3%, the transaction delay was only 21.8 ms, the service quality satisfaction and security compliance rate were 96.7% and 98.2% respectively, the double-spend attack resistance rate and resource status traceability accuracy rate both reached 99.9%, and all related indicators were significantly superior to the existing comparison schemes. This study provided technical support for 5G resource collaboration in scenarios such as industrial internet and vehicle networking, effectively solving the trust bottleneck and scheduling problems in distributed environments. However, the research has not fully considered the adaptability of resource scheduling in extreme network environments. The computational power consumption of the algorithm in large-scale node deployment scenarios must be optimized further. The computational cost of the blockchain and multi-agent deep reinforcement learning components is high. Additionally, the system’s scalability in ultra-dense 5G scenarios must be improved. To a certain extent, this framework’s immediate large-scale practical application in complex 5G network environments is limited.