基于线性编码的安全高效边缘协同计算关键技术研究

边缘计算在减少带宽消耗、云计算负载、处理时延及提高数据安全性等方面均具有优势。边缘协同计算研究已成为未来信息技术发展的重要前沿课题之一。线性编码理论已被证明可用于提高传输、存储及计算的效率和安全性，但如何将线性编码与异构、动态的边缘协同计算相结合，提高计算安全性及资源利用率尚需深入、系统的研究。本课题拟聚焦智能算法中的关键基础计算构件——大规模矩阵乘法，针对任务分配和编码计算联合优化，基于线性编码代数特性的安全、高效编码计算方案设计，以及机密性、隐私性、完整性和可靠性联合研究和统一方案设计，三个关键科学问题展开研究，设计安全、高效的边缘协同计算方案。本课题将线性编码与边缘协同计算相结合，旨在发挥线性编码的优势，充分利用广泛分布的异构边缘计算设备的通信、存储和计算资源，提高计算效率和安全性。因此，本课题研究成果不但将丰富基于编码的协同计算的理论基础，而且对边缘计算的广泛应用起到重要推动意义。

Since edge computing has considerable advantages of reducing bandwidth consumption and cloud computing load, processing delay and improving data security, edge collaborative computing has become an important research topic. Linear coding has been widely used to improve the efficiency and security of data transmission, storage and computation. However, how to employ linear coding theory to design efficient and secure edge collaborative computing still remains to be thoroughly researched. Focusing on the key computation component in intelligence algorithm, i.e., large scale matrix multiplication, this proposal will explore three key research issues, including joint optimization of task allocation and coded computing, secure and efficient coded computing design based on the algebraic properties of linear coding, and joint research and unified collaborative computing design for confidentiality, privacy, integrity and reliability. Our research will integrate the linear coding and edge collaborative computing in order to give full play to the advantages of linear coding and the communication, storage, and computing capabilities of a large number of distributed edge devices to further improve computing efficiency and security. Therefore, the expected achievements to be obtained in this research will not only enrich the theory of coded collaborative computing, but strongly promote the wide application of edge computing.