新息序列驱动的无人机控制系统数据攻击检测
Data attack detection for an unmanned aerial vehicle control system using innovation sequences

伴随物联网和自主系统的不断发展, 信息物理系统的网络安全备受关注. 无人机是一种典型的依靠通信和 控制系统实现自主飞行的智能装置, 其安全性尤为突出. 本文针对无人机的状态估计算法, 考虑其传感器和控制指 令受到数据攻击, 提出基于扩展卡尔曼滤波的新息序列状态估计检测方法. 首先建立无人机信息物理模型, 引入状 态估计算法和数据攻击模型. 然后, 利用新息序列构造标量检测统计量用于数据攻击检测, 并针对飞行器机动造成 的状态跳变引入负无穷范数, 用以降低数据攻击检测的误检率. 最后, 通过仿真实验验证所提出的检测方法能有效 检测不同威胁模式下和状态下无人控制系统的数据攻击.
With rapid advances in the fields of the Internet of things and autonomous systems, the network security of cyber-physical systems has attracted considerable attention. An unmanned aerial vehicle (UAV) is an intelligent device that relies on information communication and flight control systems to achieve autonomous flight. Consequently, its security is extremely important. This study proposes a new state estimation method that uses innovation sequences based on an extended Kalman filter for the detection of data attacks on a UAV. Our method can identify data attacks on tensors and control commands. First, a cyber-physical system model for a UAV is established, and the state estimation algorithm and data attack model are introduced. Then, a scalar detection statistic is constructed for data attack detection using innovation sequences, and the minus infinity norm method is introduced to reduce the false detection of data attacks while the aircraft is being maneuvered. Finally, simulation results show that the proposed detection method can effectively detect various threat patterns under a variety of circumstances for unmanned control systems.