Randomness of key and block features. Since we assume Eve has complete information of the pro- tocol, any non-random behavior in the bit sequences or block features can be exploited by the adversary to reduce the time-complexity of cracking the key. Although Xxxxx can generate an arbitrary key, the key may also be slightly changed after privacy amplification. We em- ploy a widely used randomness test suit, NIST to verify the randomness of the secret-bit generated by TDS. In this test, we use 200 bit sequences generated from our experiments in scenarios A, B, C, and D, and compute their p-values for 8 types of tests. According to the specification in this suite, if all p-values are greater than 0.05, the sequence is random. distance between two devices (Xxxxx and another receiver of the key). We find that when the distance is smaller than 4cm, the bit generation rate is always higher than 100 bit- s/sec. Hence it only takes a couple of seconds to get a 256-bit key. The bit generation rate in mobile scenarios is higher than that in static scenarios. The bit generation rate in outdoor environments is higher than that in indoor environments. It is because mobile and outdoor environ- ments provide more channel diversity. Compared with an- other method ProxiMate [15] that only generate a few bits per second, the bit generation rate of TDS is higher by more than an order of magnitude. Figure 14(b) shows the bit error rate by varying the dis- tance between devices, for ProxiMate and TDS. Even if the distance of two device antennas is 1cm, the bit error rate of ProxiMate is about 5%-10%. For TDS, when the dis- tance is less than 3cm, the mismatch rate of TDS is 0 for A B C 200 0.6 ProxiMate(A) TDS(A) 150 ∆σ^0 ∆σ^1 ∆σ^'0 ∆σ^'1
Randomness of key and block features. Since we assume Xxx has complete information of the pro- tocol, any non-random behavior in the bit sequences or block features can be exploited by the adversary to reduce the time-complexity of cracking the key. Although Xxxxx can generate an arbitrary key, the key may also be slightly changed after privacy amplification. We em- ploy a widely used randomness test suit, NIST to verify the randomness of the secret-bit generated by TDS. In this test, we use 200 bit sequences generated from our experiments in scenarios A, B, C, and D, and compute their p-values for 8 types of tests. According to the specification in this suite, if all p-values are greater than 0.05, the sequence is random. We list the p-values of TDS in Table 2. From the results, distance of two device antennas is 1cm, the bit error rate of ProxiMate is about 5%-10%. For TDS, when the dis- tance is less than 3cm, the mismatch rate of TDS is 0 for outdoor environments and < 0.015 for indoor environments. When the distance is 5cm, the mismatch rate of TDS is still smaller than 7%. We mark the authenticate distance and safe distance in the figure. Here the safe distance can be set to 12.5cm but a user can easily check a much longer safe distance such as 25cm or even 50cm. Out side of the safe dis- tance, a device has bit error rate equal to 0.5, the maximum bit error rate. Figure 14(c) shows the parity check counts with increas- ing the distance between devices, for ProxiMate and TDS. The number of passes is 5. When the distance is more than 1cm, parity check counts of ProxiMate are larger than 130, which might not work properly. For TDS, as long as the dis- tance is less than 5cm, the parity check counts are less than 20 in both indoor and outdoor scenarios. The devices with- in 5cm can achieve pairing without user intervention. For large civilian or military transceivers, we may use external antennas which can be easily placed in 5cm. A B C D 200 ProxiMate(A) TDS(A) 150 Bit−rate (bits/sec) 150 100 50 0 8 9 10 0.6 Bit error rate 0.5 0 ProxiMate(B) TDS(B) authen. distance safe distance 100 Parity check counts 50 Δσ^0 Δσ^1 Δσ^'0 Δσ^'1
