Privacy-preserving quantized average consensus: a general state decomposition method

This article investigates the privacy-preserving quantized average consensus problem in multi-agent systems. That is, a distributed consensus algorithm is desired to enable each node to calculate the average value of all data under quantized communication, while protecting the privacy of each node's data. This algorithm has important applications in many fields, such as distributed computing, data mining, and blockchain. To achieve this objective, a more general state decomposition method is proposed and integrated into the dynamic encoder-decoder quantization scheme to guarantee privacy-preserving accurate consensus. In this new decomposition mechanism, the number of virtual nodes generated by each node is flexible and is not determined by the network topology, but can be decomposed according to their own needs and resources, which provides a greater design freedom. Furthermore, it is strictly proved that the novel privacy-preserving quantized average consensus algorithm can achieve accurate convergence and privacy protection against both curious nodes and external eavesdroppers. Comparisons are given to show the advantages of the proposed method and obtained results. Finally, numerical simulations intuitively verify the correctness and effectiveness of main theoretical results.