Encryption-Decryption-Based Particle Filtering for Stochastic Systems With Randomly Switching Nonlinearities and Sensor Resolutions

In this paper, the secure particle filtering problem is investigated for a class of stochastic nonlinear systems subject to non-Gaussian noises and randomly switching nonlinearities. As an essential characteristic of real-world sensors, the sensor resolution is incorporated into the measurement model to provide a realistic representation of the available data. By resorting to the exclusive or logical operations, an encryption-decryption-based scheme is leveraged to enhance the transmission security of measurements and lower the communication overhead. The objective of this paper is to design a novel particle filtering scheme in the coexistence of randomly switching nonlinearities, non-Gaussian noises, sensor resolution effects and decrypted measurements. Specifically, a mixture distribution, employing the statistical property of the randomly switching nonlinearities, is constructed to generate the new particles. By considering the effects of sensor resolutions and decryption errors, the likelihood function is parameterized to facilitate the update of weights. Finally, a numerical example with Monte Carlo simulations is presented to illustrate the effectiveness of the proposed filtering algorithm.