Chen Hao, Hong Xiayun, Cheng Yao, Wang Xiaotong, Chen Liyan, Cheng Xuan, Lin Juncong
School of Informatics, Xiamen University, Xiamen, 361005, People's Republic of China.
China Mobile (Hangzhou) Information Technology Co., Ltd., Hangzhou, 311121, People's Republic of China.
Sci Rep. 2025 Apr 3;15(1):11411. doi: 10.1038/s41598-025-96350-7.
With the popularity of mobile health monitoring and genome sequencing techniques, the scale of biomedical and genomic data grow rapidly, with their privacy receiving more and more concerns. Encryption technique plays an important role in many aspects of security guarantee for these data. For mobile devices, encryption is even more important yet more challenging, as these devices are usually used in environments that may not be well protected and have rather limited computing resources. With heterogeneous multi-core processors becoming popular on mobile devices to satisfy different needs of applications, designing heterogeneous algorithms to harness all the available resources are tricky but have the potential to deliver high performance. In this research, we study how to design the heterogeneous version for AES algorithm, a representative encryption algorithm, on such processor to improve throughput and energy efficiency. To alleviate the overhead, we proposed a hybrid strategy to firstly find optimal workload allocations for cores of the processor in the offline stage and then dynamically adjust the balance in the online stage to match the running environment. We do a series of experiments on common genome data, with results showing 25-400% improvements in throughput, and 5.5-2800% improvements in energy efficiency.
随着移动健康监测和基因组测序技术的普及，生物医学和基因组数据规模迅速增长，其隐私问题也越来越受到关注。加密技术在这些数据的安全保障诸多方面发挥着重要作用。对于移动设备而言，加密更为重要但也更具挑战性，因为这些设备通常在保护欠佳且计算资源相当有限的环境中使用。随着异构多核处理器在移动设备上流行以满足不同应用需求，设计利用所有可用资源的异构算法颇具技巧性，但有潜力实现高性能。在本研究中，我们探讨如何在这类处理器上为代表性加密算法AES算法设计异构版本，以提高吞吐量和能源效率。为减轻开销，我们提出一种混合策略，首先在离线阶段为处理器的核心找到最优工作负载分配，然后在在线阶段动态调整平衡以匹配运行环境。我们对常见基因组数据进行了一系列实验，结果表明吞吐量提高了25% - 400%，能源效率提高了5.5% - 2800%。
