低成本压缩感知加密及其在无线体域网中的应用研究

The compressed sensing (CS) based cryptosystem which can perform integrative signal sampling, compression and encryption, has drawn much attention in recent years. For wireless body area network (WBAN), such a design can significantly reduce the power consumption of sensor nodes and further prolong the battery life. In literatures, CS encryption can be divided into two classes, with the first one uses secret measurement matrix, whereas other ciphers are introduced to re-encrypt the measurements in the second class. However, the former is vulnerable against plaintext attack, while the latter is more power-consuming. Aiming to develop low-cost CS encryption, the following tasks will be conducted in this project: 1) novel CS architecture whose measuring matrix doesn’t own restricted isometry property, with the purpose to ensure the security even the measurement matrix is exposed; 2) a quantization approach that has diffusion effect and can conceal the signal’s energy, so as to introduce nonlinear secrecy into the quantization process; 3) suitable mode of operation for CS encryption, to introduce probability into the deterministic encryption process; 4) implementation of the above approaches for bio-signal compressive sampling and secure transmission, for comprehensively evaluation of the proposed algorithms. By reforming the traditional viewpoints which focus on secret sampling matrix and measurements re-encryption, this project significantly broaden the research horizon of CS cryptography. Besides, the research achievements are essentially solutions for simultaneous signal sampling, compression and encryption, and hence very beneficial for WBAN and other low-end internet of things applications.

压缩感知加密算法能实现一体化的信号采样、压缩和加密，近年来备受关注。对于无线体域网而言，这将显著降低传感器的功耗进而延长续航时间。当前，压缩感知加密算法主要有两类：一是使用秘密的测量矩阵，二是引入其他密码再次加密测量值。但是，前者无法抵抗明文攻击，安全性能不足；而后者会耗费更多资源，成本较高。本项目以低成本压缩感知加密为目标，开展以下研究：1）测量矩阵不满足有限等距性质的压缩感知，确保测量矩阵暴露时依然安全；2）具备扩散效果并掩盖信号能量的量化方法，在量化中嵌入非线性密码特征；3）适用的工作模式，在确定性加密过程中引入随机特性；4）将上述成果用于生理信号的压缩采样和加密传输，全面评估算法性能。本项目突破以秘密测量矩阵和测量值再次加密为核心的传统思路，不断拓宽压缩感知加密的研究视野；相关成果作为采样、压缩、加密有机融合的一体化解决方案，在无线体域网等终端低成本的物联网应用中有重要的实用价值。

当前，以无线体域网(Wireless Body Area Network, WBAN)为代表的新型医疗健康服务模式快速发展，拥有广泛的应用领域和广阔的发展前景。在此过程中，信号的压缩采样和数据加密对提高系统续航能力和保护用户隐私至关重要。然而，传统的香农采样-数据压缩-信息加密思路需要较高的计算资源和能量消耗，对硬件资源极其受限的 WBAN 并不适合。本项目探讨使用压缩感知(Compressed Sensing, CS)作为解决方案。在确保CS压缩采样特征的基础上，设法将数据加密有机嵌入信号采样过程，构建采样、压缩、加密融为一体的压缩感知体制，为WBAN提供低功耗、高机密性的信号采样方案，从而降低节点的资源需求、延长监控时间，不断提升WBAN的实际应用潜力。取得了几个方面的成果，一是探讨压缩感知在无线体域网中的应用潜力，开发一体化的信号采样、压缩和加密方案，提出并验证了基于压缩感知的多个低成本心电心音压缩采样方案。二是发掘压缩感知的保密特性，对当前经典的置乱替换架构及融合压缩感知的加密算法进行安全分析，并提出安全高效的压缩感知加密方案。相关研究内容按计划圆满完成。在本项目的资助下，项目负责人以第一作者或通信作者发表学术论文15篇，其中SCI检索期刊论文14篇；受邀在国内外学术会议和科研院所发表讲座十余次，获得辽宁省科技进步二等奖一项。此外，在相关项目成果的支撑下，进一步的拓展工作获国家自然科学基金面上项目资助。同时，通过本项目的实施，项目组成员得到了极大的科研锻炼。