多媒體網路安全實驗室 Mobility Assisted Secret Key Generation Using Wireless Link Signatures Date： Reporter : Hong Ji Wei Auther : Junxing Zhang Kasera, S.K. Patwari, N. 出處 : INFOCOM, 2010 Proceedings IEEEINFOCOM, 2010 Proceedings IEEE

多媒體網路安全實驗室 Outline INTRODUCTION 1 ADVERSARY MODEL 2 MOBILITY ASSISTED KEY ESTABLISHMENT 33 PROTOCOL EVALUATION 44 CONCLUTIONS 35

多媒體網路安全實驗室 INTRODUCTION  these link signatures can be measured almost symmetrically between two ends of a wireless link.  location locking attack:the adversary steals some signature measurements it has a good chance to determine the key generated.  CIR:the channel impulse response

多媒體網路安全實驗室  the wireless link signatures at different unpredictable locations and combine these measurements to produce strong secret keys.  Using extensive measurements in both indoor and outdoor settings  (i) when movement step size is larger than one foot the measured CIRs are mostly uncorrelated  (ii) more diffusion in the mobility results in less correlation in the measured CIRs

多媒體網路安全實驗室 ADVERSARY MODEL  an adversary:that can overhear all the communication between the two devices A and B.  Assume that the adversary cannot cause a person-in-the-middle attack.  Our adversary is also not interested in causing any Denial-of-Service attacks

多媒體網路安全實驗室 MOBILITY ASSISTED KEY ESTABLISHM ENT A.Key Establishment Protocol Phase1:SIGGEN (short for signature generation)  A and B exchange SIGGEN and SIGACK messages.  Between each pair of SIGGEN and SIGACK message exchange, A and B individually, or both move to a new location.

多媒體網路安全實驗室 Phase2:SIGCHK (short for signature check)  Upon receiving the SIGCHK message from A, B quantizes all CIR it has measured and removes any duplicates.  then encodes the remaining quantized CIRs to produce both message symbols and parity symbols.  B sends only the parity symbols to A in multiple SIGFEC (short for signature forward error correction) messages.  A quantizes the corresponding CIRs that she had measured and encodes them to produce message symbols.

多媒體網路安全實驗室  A then combines her message symbols with parity symbols she receives from B to obtain a bit stream that is identical to that of B.  In the final KEYGEN (short for key generation) phase, A and B generate a new secret key with the reconciled bit streams and verify.  To convert the bit stream obtained: utilize a key compression function(SHA-256, SHA-384, and SHA-512).

多媒體網路安全實驗室 B. Quantization and Bit Extraction  Because CIRs are continuous random variables, must quantize them in order to use them for secret key generation.  first normalize each CIR with its maximum element value.  Next, to quantize the normalized CIR to 2 q discrete values with equal intervals.  simply convert integers in the resulting vector to their binary representation to extract the initial bits.

多媒體網路安全實驗室  C. Jigsaw Encoding  the simple uniform quantization cannot preserve reciprocity and even increase the discrepancy rate in quantized CIRs.

多媒體網路安全實驗室

D. RS Error Correction  adopt the RS forward error correction (FEC) scheme  Each RS output codeword has p symbols including k input symbols followed by 2 × t parity symbols.  t :the errorcorrection capability  ε:the link signature discrepancy rate.

多媒體網路安全實驗室  the computational complexity Γ  EX: 1.For m = 10 and q = 5, it is larger than For m = 10 and q = 1, 2, it is in the order of 2 427

多媒體網路安全實驗室 PROTOCOL EVALUATION  A. Measurement Campaign  we use three mobility models: random walk, Levy walk, and Brownian motion  Levy walk:  Brownian motion: 懸浮在液體或氣體中的微粒所 作的永不停息的無規則運動

多媒體網路安全實驗室  Impact of Mobility on Link Signatures

多媒體網路安全實驗室

C.Quality of Key Generation

多媒體網路安全實驗室  we use a metric called Secret Bit Rate that is defined as the average number of secret bits extracted from each channel response.  We plot the entropy values of the bit stream generated with different quantization bit numbers (per channel response).

多媒體網路安全實驗室

CONCLUTIONS  We propose an approach where wireless devices,interested in establishing a secret key.  Our results show that our scheme generates very high entropy secret bits and that too at a high bit rate.

多媒體網路安全實驗室