支持数据实用性和容错的差分隐私保护方案

针对智能电网环境下个体数据的差分隐私与聚合数据实用性的均衡问题，提出基于近似耗电分组的差分隐私算法，通过降低组内耗电值的最大敏感度，降低整体差分隐私噪音，提高聚合数据对于供电方的实用性；针对内部节点攻击个体电表数据的问题，通过构建分布式加密聚合平台，抵御包括控制中心在内的内部节点对个体细粒度数据的攻击；解决由于故障电表的存在所导致的分布式聚合方案不能正确解密同态加密聚合值以及非故障电表添加的噪音值不能满足整体差分噪音量需求这2个问题. 实验证明所提出的基于近似耗电分组的算法与预估故障率设定差分噪音的方法的结合，相比其他相近方案，在提高聚合数据实用性方面有明显提升，同时分布式加密聚合平台为抵御内部节点攻击以及支持加密容错和差分容错提供了轻量级保证.

Differential privacy protection scheme supporting high data utility and fault tolerance

Aiming at the problem of balancing the differential privacy of individual data and the aggregation data utility under the smart grid environment, a differential privacy algorithm based on similar power consumption grouping was proposed. By reducing the maximum sensitivity of consumption data, the whole differential privacy noise was reduced, and the utility of aggregation data for the power supplier was improved. To solve the problem of internal nodes attacking individual data, a distributed encryption aggregation platform was constructed to resist the attack of internal nodes including the control center on individual fine-grained data. The proposed method can solve the two issues due to the existence of the malfunctional smart meters, i.e. the distributed aggregation scheme cannot correctly decrypt the homomorphic encryption aggregation data and the added noise of the non-malfunctional smart meters cannot satisfy the overall differential requirement. Experiments show that the combination of the proposed method based on similar consumption grouping and the method of estimating the failure rate and setting the differential noise, compared with other related schemes, has an obvious effect on improving the utility of aggregation data, and the distributed encryption aggregation platform also provides lightweight guarantee for resisting the attack of internal nodes as well as supporting the encrypted fault-tolerance and the differential fault-tolerance.