天基智能模型的协同可信分布式推理技术

Collaborative and Trustworthy Space-Based Distributed Inference Technology for Intelligent Models

  • 摘要: 针对天基网络中多节点分布式推理及推理过程中的安全隐私挑战,研究协同可信的天基智能模型分布式推理技术. 构建基于区块链的天基智能模型推理系统——包含云端节点和边缘卫星节点集群,提出协同可信的天基智能模型分布式推理方法,准备阶段基于负载均衡和时延最小对模型进行自适应分割,推理阶段采用区块链技术实现协同可信的集群推理. 实验结果表明,所设计的系统及方法能够保障分布式节点安全有效地完成智能模型分布式推理,为智能模型在卫星场景中的应用提供支撑.

     

    Abstract: In response to the security and privacy challenges of multi-node distributed inference in satellite networks, a collaborative and trustworthy distributed inference technology for space-based intelligent models was investigated. Firstly, a space-based intelligent model inference system was constructed based on leveraging block chain technology, considering cloud nodes and clusters of edge satellite nodes. And then, a method was proposed for the collaborative and trustworthy distributed inference of space-based intelligent models. In the preparation stage, an adaptive segmentation for the models was performed based on load balancing and minimal latency considerations. During the inference stage, a blockchain technology was adopted to realize the collaborative and credible cluster inference. Experimental results demonstrate that the proposed system and method can achieve the safe and efficient distributed inference of intelligent models, thereby supporting the application of intelligent models in satellite scenarios.

     

/

返回文章
返回
Baidu
map