Secure Decision Fusion in ISAC-Oriented Distributed Wireless Sensing Networks with Local Multilevel Quantization

Abstract

Distributed deployment for integrated sensing and communication (ISAC) can improve the sensing accuracy by exploring spatial diversity for covering the target state. However, secure fusion and limited energy consumption are still challenges for wireless-transmission-based distributed ISAC. In this paper, a secure decision-fusion scheme under energy constraint is proposed. First, the local likelihood ratios (LRs) of the local observations at sensing nodes are quantified at multiple levels corresponding to a multiple phase-shift keying (MPSK) constellation, in order to retain more sensing information. Second, an antieavesdropping scheme, which randomly rotates the constellation based on the main channel information between the nodes and ally fusion center (AFC), is proposed to confuse the data fusion of the eavesdropping fusion center (EFC). In addition, the local quantization thresholds and the rotating threshold are optimized to realize the perfect security under energy constraint and maximum rotation angle of π. In addition, the optimized rotation angle is discussed under a relaxed security requirement of the EFC in exchange for reducing the AFC error. Performance evaluation results show that the AFC error probability of the proposed scheme with a two-bit quantization and soft fusion outperforms the single-bit case and three-bit case by above 3 dB and about 0.5 dB at the error probability of 10−2, respectively. The former gain is just contributed by the more local information kept with two-bit against single-bit quantization. However, for the three-bit case, the advantage of more levels of quantization is eliminated by the worse transmission of denser constellation over a noisy channel. Moreover, the proposed scheme outperforms the conventional channel-aware encryption method under a stricter energy constraint and higher signal noise ratio (SNR).