Intelligent Reflecting Surface-Assisted Secret Key Generation with Discrete Phase Shifts in Static Environment

  In this section, we present results to evaluate the proposed scheme. To ensure the accuracy, we perform Monte Carlo simulation to generate 100000 random channel coefficients in every coherence time slots. Unless otherwise noted, we set σ2= σ2 n= 1, α = 1, Pt= PA= PB, and SNR=γ=γa=γbin the following simulation. 、

First, we plot 1000 channel estimation samples in Fig.2. It can be observed from Fig.2(a) that the channel coefficient follows the complex Gaussian mixture distribution when N= 5 and has KNdifferent values. Fig.2(b) and Fig.2(d) (i.e., the cumulative probability distribution) indicate that the real parts of the channel coefficients approximately follow Gaussian distribution when N= 20. In addition, it can be observed from Fig.2(b) that the channel coefficients of Alice and Bob are very similar, but they are quite different from Eve’s in Fig.2(c), which proves that the eavesdropping channel and the legitimate channel are uncorrelated as [1]–[3] and [9].