Comparison of key extraction approaches Clause Samples

Comparison of key extraction approaches. We compare TDS with existing key generation and agree- ment approaches for mobile networks, including KEEP [28], ▇▇▇▇▇▇ et al. [16], ASBG [8], CGC [12]. Note these solu- tions assume an authenticated channel between two devices. Hence they are weaker in security than TDS. We align the baseline of comparison as follows. In the scheme proposed by ▇▇▇▇▇▇ et al., there are two parameters α and m. We set α = 0.35 and m = 2 to ensure most fractions of mea- surements are used for bit extraction. For ASBG, CGC, and KEEP, we choose α = 0.35 and fragment size is 50, where the mismatch ratio is low. For TDS, we choose block size β = 6 in static scenarios and β = 4 in mobile scenarios. The distance between ▇▇▇▇▇ and another device is within 4cm. We compare the entropy of keys generated by different ap- proaches in Figure 17. The entropy can reflect the random- Entropy 0 A B C D 0.04 TDS KEEP ASBG Mathur CGC Mismatch rate 0.03 0.02 0.01 0 B C D 120 Secret bit rate (bits/sec) 90 60 30 0 A B C D 40 Information reconcilation counts 30 20 10 0 A B C D Scenarios Scenarios Scenarios ▇▇▇▇ of keys from the perspective of uncertainty. TDS and KEEP have the highest entropy in all methods, and CGC has the lowest. Figure 18 shows the bit error rates. In this distance, TDS has no mismatched bit, while other methods may cause around 2% to 4% mismatched bits. Figure 19 shows the bit generation rates. Obviously TDS has signifi- cantly higher generation speed. Note the bit generation rate of TDS is slower than previous results. It is because in this set of experiments, ▇▇▇▇▇ and ▇▇▇ do not listen to a public WiFi but use the communication among them for sampling. This is the only model that the other protocols can work but TDS is not restricted to it. Figure 20 shows the num- ber of rounds for information reconciliation. Since there are no mismatched bit, TDS only uses 4 times pass check to guarantee the consistency of transmitted secret bits.