Secret key generation using entropy-constrained-like quantization scheme
Xuanxuan Wang, Lars Thiele, Thomas Haustein, Yongming Wang · 2016
Tremendous researches have been carried out on quantization techniques for secret key generation, but there is little consideration about the causes of quantization discrepancy. In fact, the non-reciprocity channel measurements and the unavoidable quantization noise could result in the quantization discrepancy of the quantization bits at the two intended nodes, which has an important effect on the quantizer design. Moreover, these studies ignore the fact that the quantization level should be bounded by the mutual information between the two intended nodes. In this paper, we first analyze the factors of quantization discrepancy and then propose the Entropy-Constrained-like Quantization Scheme (ECQS) to minimize the quantization discrepancy. This scheme mainly focuses on two segments: non-reciprocity balancing and Entropy-Constrained-like optimal quantization. Simulation results show that, with this scheme, the non-reciprocity channel components can be well balanced and the bit mismatch ratio (BMR) is reduced significantly.