CN103825725A - Efficient random physical layer secrete key generation method based on vector quantization - Google Patents

Abstract

The invention discloses an efficient random physical layer secrete key generation method based on vector quantization. According to the method, a first wireless network communication party and a second wireless network communication party respectively and simultaneously detect RSS information of a wireless channel and transmit the information to each other, and the first wireless network communication party and the second wireless network communication party operate in the steps of inconsistent removal, vector quantization and fuzzy extraction and ultimately obtain identical bit streams to be used as encrypted secrete keys. The bit generation rate can reach 284%, and random bit streams can be quickly provided for authentication encrypted algorithms of the two communication parties in a wireless network environment to serve as the encrypted secrete keys; the zero error rate can be achieved through the two communication parties, it can be guaranteed that the two communication parties reliably generate the identical bit streams to be used as the encrypted secrete keys, and stability of the method in the operation process is guaranteed; the bit streams produced in the method can pass an NIST randomness test, the produced random bit streams can be used as the secrete keys, and safe communication of the two communication parties is guaranteed.

Description

A kind of efficient random physical layer key generation method based on vector quantization

Technical field

This invention belongs to wireless network secure field, relates to a kind ofly efficiently generating randomly method and the relevant apparatus for the safe key of cryptographic algorithm from physical layer.In general, this invention can meet the generation speed of wireless network secure field to safe key and the requirement of security intensity.

Background technology

Various wireless devices have become a part indispensable in our daily life, and by these equipment, we can carry out information sharing and exchanges data.For example, one group of visitor need to carry out the resource such as image sharing, video by point-to-point connection, prevents that again others from visiting these resources simultaneously.In order to dispose safely and adopt these application, guarantee that confidentiality, integrality and the authentication property of transfer of data just seems particularly important.

The broadcast characteristic of wireless network makes it face the safety problem larger than cable network, and assailant can easier eavesdropping or intercepted the Content of Communication of legal node, and wireless device can be modified or be counterfeiting to carry out some malicious acts.

Existing wireless encryption system exists many weak points in some scene, specific as follows:

(1) existing 802.11 safe practices provide authentication and privacy mechanism for Frame, but management frames and control frame are not protected.

(2) theoretical foundation of existing safety system is mathematical difficult problem (as large Integer Decomposition problem and discrete logarithm computational problem).Along with the appearance of the development of computing technique, particularly quantum computer, assailant's computing capability significantly promotes, and the fail safe of these cryptographic techniques based on computational complexity will reduce.

(3) most cryptographic technique, particularly public key cryptography have higher complexity and calculation requirement, so be not suitable for being applied in low distributed wireless networks of joining.

(4) traditional security mechanism based on shared key needs key management, distribution, renewal and maintenance, and along with the increase of node number, required key number is exponential increase, and the workload of key distribution and renewal can be very large.And in Ad hoc network, node frequently adds and deviated from network, this makes key management more difficult.

Based on above reason, the scheme that introducing physical layer strengthens safely wireless network secure has received increasing concern.Physical layer key is based on Information theoretical secure, and the difficulty that it can reduce encryption key distribution and renewal, realizes layer security, improves the fail safe of existing key code system.

The scheme that produces key from physical layer need to be take physical layer channel information as basis.These information comprise angle of arrival degree, phase place, received signal strength (RSS), channel impulse response (CIR), signal envelope and cross-point etc.Wherein, received signal strength (RSS) is the common method that generates key, and this is because RSS can directly be obtained by existing network interface card.

In general, the standard of measurement physical layer key generation method quality comprises following three kinds:

(1) bit production rate: the final key bit number generating of single RSS;

(2) the key error rate: communicating pair generates in key and do not mate the figure place of key and the ratio of key total length;

(3) key randomness: 0 and 1 uniformity coefficient distributing in the binary stream of generation key.

Current, all there is identical problem in the physical layer key generation method that wireless network secure field exists, and these methods all can not meet the requirement of higher bit production rate, low error rate, high randomness simultaneously.They mostly sacrifice bit production rate to reduce the error rate, and inefficient key generation method causes node to need lot of energy to carry out repeatedly sounding channel to produce the key of sufficient length.

Summary of the invention

In view of the deficiencies in the prior art, the present invention aims to provide a kind of efficient random physical layer key generation method based on vector quantization, produces rapidly random bit stream and encrypts and authentication provides reliable key for communicating pair under wireless environment.

To achieve these goals, the technical solution used in the present invention is as follows:

A kind of efficient random physical layer key generation method based on vector quantization, the wireless communication both sides RSS information of detection wireless channel be transferred to the other side simultaneously respectively, wireless communication both sides take inconsistency removal, vector quantization and fuzzy extraction step to operate simultaneously, finally obtain identical bit stream and use as encryption key.

Described method comprises the steps:

Step 1, inconsistency are removed: remove the inconsistent channel information causing due to transmission channel half-duplex or ambient noise;

Step 2, vector quantization: RSS is divided into two intervals according to average line, finally discontinuous RSS value is converted into 0,1 bit stream;

Step 3, fuzzy extraction: wireless communication both sides utilize Fuzzy extractor to carry out error correction and export identical bit stream as encryption key generating inconsistent position in bit stream.

Further, inconsistency removal stage (step 1) comprises the steps:

Step 1.1, input: the RSS array X={x collecting ₁, x ₂..., x _k, x _i∈ Z;

Step 1.2, piecemeal: RSS value is carried out to piecemeal, piece is called to block, k RSS value will belong in different block, and making the length of each block is b, suppose total t block;

Step 1.3, use label function R (x _i) RSS value is quantized, the RSS value not being quantized is abandoned;

Step 1.4, definition m are improvement factor, wireless communication Party A according to label function generate 0,1 sequence, run into continuous m position identical 0 or 1, the sequence number of interposition is recorded in to L _{a_to_b}in array, generate L _{a_to_b}={ l ₁, l ₂..., l _a, and send to wireless communication Party B;

Step 1.5, wireless communication Party B generate 0,1 sequence according to label function, run into continuous m position identical 0 or 1, the sequence number of interposition is recorded in to L _{b_to_a}in array, and and L _{a_to_b}array contrasts, and rejects inconsistent sequence number, generates new L _{b_to_a}array also sends to wireless communication Party A.

It should be noted that described label function R (x _i) be:

Wherein, with reference to boundary array be $Q+=\{q_1^{+},q_2^{+},...,q_t^{+}\}$ With $Q-=\{q_1^{-},q_2^{-},...,q_t^{-}\},$ $q_i^{+}={\mathrm{mean}}_i+\mathrm{\α}*\mathrm{std}\_{\mathrm{derivation}}_i,q_i^{-}={\mathrm{mean}}_i-\mathrm{\α}*\mathrm{std}\_{\mathrm{derivation}}_i,$ Mean _ibe the RSS mean value of i block, α is the fluctuation factor and α ∈ (0,1), std_derivation _iit is the RSS standard deviation of i block.

Further, (step 2) comprises the steps: the vector quantization stage

Step 2.1, input through inconsistency remove after effective RSS array Y and the dimension N of vector quantization;

Step 2.2, for the each y in array Y _iall set up N n dimensional vector n;

Step 2.3, to each y _iall apply N Dimension Vector Quantization of Linear Prediction device Q _n(y _i), 0,1 bit stream producing after quantizing is as output.

It should be noted that, the RSS array Y in described step 2.1 is: Y={y ₁, y ₂..., y _d, y _i∈ Z, the RSS value in Y array is L _{b_to_a}the corresponding RSS value of each sequence number in array.

It should be noted that, be y in described step 2.2 _ithe N n dimensional vector n of setting up is: $<y_i,y_{(i+{\mathrm{\&Delta;}}_1)\mathrm{mod}d},y_{(i+{\mathrm{\&Delta;}}_1+{\mathrm{\&Delta;}}_2)\mathrm{mod}d},...,y_{(i+{\mathrm{\&Delta;}}_1+...+{\mathrm{\&Delta;}}_{N-1})\mathrm{mod}d}>$ Wherein, Δ={ Δ ₁, Δ ₂..., Δ _n-1the component interval of N component in N n dimensional vector n, Δ _jit is the sequence number interval between j component and j+1 component.

Need to further illustrate the N Dimension Vector Quantization of Linear Prediction device Q in described step 2.3 _n(y _i) be: $Q_N\left(y_i\right)=R\left(y_i\right)R\left(y_{(i+{\mathrm{\&Delta;}}_1)\mathrm{mod}d}\right)...R\left(y_{(i+{\mathrm{\&Delta;}}_1+...+{\mathrm{\&Delta;}}_{N-1})\mathrm{mod}d}\right),$ Wherein, y _ifor the RSS value of N Dimension Vector Quantization of Linear Prediction device input.

In fuzzy leaching process (step 3), Fuzzy extractor can be defined as: <Gen, and Rep>, parameter is (M, l, t), wherein, process Gen is probability generative process, process Rep is certainty recovery process, M is incoming bit stream set, and l is incoming bit stream length, and t is the ultimate range restriction between wireless communication Party A and the message of wireless communication Party B input.

First described process Gen is operated by wireless communication Party A, then the parameter P of output is passed to wireless communication Party B, wireless communication Party B is again according to process Rep operation, and wireless communication Party A and wireless communication Party B produce on all four random secret information R as encryption key.

Process Gen is: (R, P) ← Gen (w ₀), wherein, input parameter w ₀for the bit stream that length is l, output parameter R is that length is the random secret information of l, and output parameter P is public information.

Process Rep is: R ← Rep (w', P), wherein, input parameter w' is that length is l and and w ₀distance be not more than the bit stream of t, input parameter P is the public information for error correction that process Gen generates, output parameter R is that length is the random secret information identical with Gen process of l.

It should be noted that, process Gen can preferentially adopt and realizes with the following method but be not limited to following method: utilize Bose-Chaudhuri Hocquenghem error correction codes to w ₀just error correction, the error correcting code of generation, as P, utilizes SHA-1 algorithm to w ₀carry out Hash processing, the input of Hash is 1:1 with output Length Ratio, generates the strong key of randomness as R.

It should be noted that, process Rep can preferentially adopt and realizes with the following method but be not limited to following method: utilize the error correction sequence P producing in Bose-Chaudhuri Hocquenghem error correction codes and Gen process to carry out error correction to w', recover w ₀sequence, utilizes SHA-1 algorithm to w ₀carry out Hash processing, the input of Hash is 1:1 with output Length Ratio, generates the strong key of randomness as R.

Beneficial effect of the present invention is:

(1) bit production rate of the present invention can reach 284%, and average each RSS value can produce 2.84 bits, can be that the authentication encryption algorithm Quick of communicating pair under wireless network environment supplies random bit stream as encryption key;

(2) the present invention is in application, and communicating pair can reach zero error rate, can guarantee that communicating pair generates on all four bit stream reliably as encryption key, has guaranteed the stability of the present invention in utilization process;

(3) bit stream that the present invention produces has passed through NIST randomness test, can provide random bit stream as encryption key for authentication encryption algorithm, has guaranteed the secure communication of communicating pair.

Accompanying drawing explanation

Fig. 1 is schematic flow sheet of the present invention;

Fig. 2 is that two-dimensional vector quantizes schematic diagram;

Fig. 3 is the organigram of Fuzzy extractor.

Embodiment

Below in conjunction with drawings and Examples, the invention will be further described.

As shown in Figure 1, the present invention is the efficient random physical layer key generation method of Replication Approach in Mobile Environment based on vector quantization, wireless communication both sides Alice and the Bob RSS information of detection wireless channel be transferred to the other side simultaneously respectively, both sides take this method to operate simultaneously, finally obtain identical bit stream and use as encryption key.

Described method comprises the steps:

Step 1, inconsistency are removed: remove the inconsistent channel information causing due to transmission channel half-duplex or ambient noise;

Step 2, vector quantization: RSS is divided into two intervals according to average line, finally discontinuous RSS value is converted into 0,1 bit stream;

Step 3, fuzzy extraction: both sides utilize Fuzzy extractor to carry out error correction and export identical bit stream as encryption key generating inconsistent position in bit stream.

Further, inconsistency removal stage (step 1) comprises the steps:

Step 1.1, input: the RSS array X={x collecting ₁, x ₂..., x _k, x _i∈ Z;

Step 1.2, piecemeal: RSS value is carried out to piecemeal, piece is called to block, k RSS value will belong in different block, and making the length of each block is b, suppose total t block;

Step 1.3, use label function R (x _i) RSS value is quantized, the RSS value not being quantized is abandoned;

Step 1.4, definition m are improvement factor, Alice according to label function generate 0,1 sequence, run into continuous m position identical 0 or 1, the sequence number of interposition is recorded in to L _{a_to_b}in array, generate L _{a_to_b}={ l ₁, l ₂..., l _a, and send to Bob;

Step 1.5, Bob generate 0,1 sequence according to label function, run into continuous m position identical 0 or 1, the sequence number of interposition is recorded in to L _{b_to_a}in array, and and L _{a_to_b}array contrasts, and rejects inconsistent sequence number, generates new L _{b_to_a}array also sends to Alice.

It should be noted that described label function R (x _i) be:

Wherein, with reference to boundary array be $Q+=\{q_1^{+},q_2^{+},...,q_t^{+}\}$ With $Q-=\{q_1^{-},q_2^{-},...,q_t^{-}\},$ $q_i^{+}={\mathrm{mean}}_i+\mathrm{\&alpha;}*\mathrm{std}\_{\mathrm{derivation}}_i,q_i^{-}={\mathrm{mean}}_i-\mathrm{\&alpha;}*\mathrm{std}\_{\mathrm{derivation}}_i,$ Mean _ibe the RSS mean value of i block, α is the fluctuation factor and α ∈ (0,1), std_derivation _iit is the RSS standard deviation of i block.

Further, (step 2) comprises the steps: the vector quantization stage

Step 2.1, input through inconsistency remove after effective RSS array Y and the dimension N of vector quantization;

Step 2.2, for the each y in array Y _iall set up N n dimensional vector n;

Step 2.3, to each y _iall apply N Dimension Vector Quantization of Linear Prediction device Q _n(y _i), 0,1 bit stream producing after quantizing is as output.

It should be noted that, the RSS array Y in described step 2.1 is: Y={y ₁, y ₂..., y _d, y _i∈ Z, the RSS value in Y array is L _{b_to_a}the corresponding RSS value of each sequence number in array.

It should be noted that, be each y in described step 2.2 _ithe N n dimensional vector n of setting up is: $<y_i,y_{(i+{\mathrm{\&Delta;}}_1)\mathrm{mod}d},y_{(i+{\mathrm{\&Delta;}}_1+{\mathrm{\&Delta;}}_2)\mathrm{mod}d},...,y_{(i+{\mathrm{\&Delta;}}_1+...+{\mathrm{\&Delta;}}_{N-1})\mathrm{mod}d}>$ Wherein, Δ={ Δ ₁, Δ ₂..., Δ _n-1the component interval of N component in N n dimensional vector n, Δ _jit is the sequence number interval between j component and j+1 component.

Need to further illustrate the N Dimension Vector Quantization of Linear Prediction device Q in described step 2.3 _n(y _i) be: $Q_N\left(y_i\right)=R\left(y_i\right)R\left(y_{(i+{\mathrm{\&Delta;}}_1)\mathrm{mod}d}\right)...R\left(y_{(i+{\mathrm{\&Delta;}}_1+...+{\mathrm{\&Delta;}}_{N-1})\mathrm{mod}d}\right),$ Wherein, y _ifor the RSS value of N Dimension Vector Quantization of Linear Prediction device input.

Two-dimensional vector quantizes as shown in Figure 2, when being input as y _itime, with Δ _ifor finding backward second component in interval

composition two-dimensional vector

and calculate label function output valve R (y according to these two components _i) and

this label output valve of two

be 01,01 to be y _itwo-dimensional vector quantize output.

As shown in Figure 3, in fuzzy leaching process (step 3), Fuzzy extractor can be defined as: <Gen, Rep>, parameter is (M, l, t), wherein, process Gen is probability generative process, and process Rep is certainty recovery process, and M is incoming bit stream set, l is incoming bit stream length, and t is the ultimate range restriction between Alice and the message of Bob input.

First described process Gen is operated by Alice, then the parameter of output is passed to Bob, and Bob is again according to process Rep operation, and Alice and Bob produce on all four random secret information R as encryption key.

Process Gen is: (R, P) ← Gen (w ₀), wherein, input parameter w ₀for the bit stream that length is l, output parameter R is that length is the random secret information of l, and output parameter P is public information.

Process Rep is: R ← Rep (w', P), wherein, input parameter w' is that length is l and and w ₀distance be not more than the bit stream of t, input parameter P is the public information for error correction that Gen process generates, output parameter R is that length is the random secret information identical with Gen process of l.

Embodiment

In the present embodiment, it is upper that communicating pair Alice and Bob operate in respectively two PC, and the operating system of lift-launch is ubuntu12.04, all uses the network interface card Atheros TL-WN650G of same model, is driven by Madwifi, is operated in transmitting-receiving bag under 802.11g pattern.In this implementation process, we revise Madwifi, have added the reply to beacon frame, beacon_ack frame carries the sequence number identical with beacon frame, by the coupling of sequence number, communicating pair can complete the pairing of RSS, and the length of beacon_ack is 49 bytes.Bob moves around and sends beacon frame take 100ms as interval to Alice around Alice, and Alice can record sequence number and corresponding RSS value after receiving beacon frame, and sends at once the acknowledgement frame of a beacon_ack to Bob.Bob receives after acknowledgement frame, records equally sequence number and corresponding RSS value.In the present embodiment, 120000 RSS have been collected altogether.

Step 1, inconsistency is removed.

Because Alice and Bob there will be some errors in the time measuring RSS value, both sides' fluctuating range is always not the same, so by the processing of RSS piecemeal, calculate and quantize bound for each piece, can make the RSS similarity of each interior Alice and Bob higher, thereby reduce the error rate.In the present embodiment, the big or small block value of RSS sequence piecemeal is 80.

Fluctuation factor-alpha is the important factor in order that reduces the error rate, and α larger (little), with reference to boundary Q ⁺and Q ^-the distance (closely) far away of range averaging line, there is disallowable just higher (low) of probability of point of little deviation, the error rate can decline (rising), but simultaneously two with reference to the point being rejected between boundary also just more (lack), can cause bit production rate decline (rising).In the present embodiment, α value is 0.2.

Improvement factor m is larger, meets continuous m position and is all greater than Q ⁺or be less than Q ^-point fewer, consult effective RSS value of generating just fewer, so can significantly reduce along with m increases bit production rate, and m increases and means the significantly situation minimizing of saltus step of continuous m position RSS, so the error rate can corresponding decline; Otherwise m reduces effectively RSS number can be increased greatly, bit production rate increases, and the significantly situation increase of saltus step of m position continuously, the error rate reduces, and in the present embodiment, m value is 2.

Step 2, vector quantization.

N Dimension Vector Quantization of Linear Prediction has represented the corresponding output that generates N bit of the input of an effective RSS, so the larger bit output of N is longer, can be larger for the bit stream that generates key, and finally make bit production rate increase.Increase but N means more greatly the number that bit is reused, the independence between bit reduces, and randomness weakens.In the present embodiment, choose Δ ₁=Δ ₂=...=Δ _n-1=60, it is more even that the vector that now each RSS is corresponding is mapped to distribution in N dimension space reference axis, and choosing of Δ all do not affect bit production rate and the error rate in addition.

Step 3, fuzzy extraction.

In the present embodiment, carry out the process of fuzzy extraction as follows:

(1) Alice generates bit stream w by vector quantization ₀, as the input of Gen, generate error correction sequence P by BCH (23,12) error correction coding, and to w ₀carry out SHA-1 Hash and generate the strong key of randomness, the input of Hash is 1:1 with output Length Ratio.

(2) error correction sequence P is sent to Bob by Alice, and the bit stream w' that Bob uses P to quantize to generate to oneself carries out error correction, recovers w ₀sequence, then carries out equally SHA-1 Hash and generates identical random key.

The former code length of BCH (23,12) of choosing in this process is 12, and error correction code length is 11, and its minimum distance is 7, can correct 3 mistakes.Bit stream is divided into every 12 one group by we, and 11 error correcting codes that generate correspondence send to the other side of communication, and these 11 error correcting codes can be corrected at most 3 bit-errors in 12 true forms.After BCH code error correction, the key that Alice and Bob generate is in full accord, does not have any error bit, reaches zero error rate.

For a person skilled in the art, can be according to technical scheme described above and design, make other various corresponding changes and distortion, and these all changes and distortion all should be within the protection ranges of the claims in the present invention.

Claims (10)

1. the efficient random physical layer key generation method based on vector quantization, it is characterized in that: the wireless communication both sides RSS information of detection wireless channel be transferred to the other side simultaneously respectively, wireless communication both sides take inconsistency removal, vector quantization and fuzzy extraction step to operate simultaneously, finally obtain identical bit stream and use as encryption key.

2. a kind of efficient random physical layer key generation method based on vector quantization according to claim 1, is characterized in that, described method comprises the steps:

Step 1, inconsistency are removed: remove the inconsistent channel information causing due to transmission channel half-duplex or ambient noise;

Step 2, vector quantization: RSS value is divided into two intervals according to average line, finally discontinuous RSS value is converted into 0,1 bit stream;

Step 3, fuzzy extraction: wireless communication both sides utilize Fuzzy extractor to carry out error correction and export identical bit stream as encryption key generating inconsistent position in bit stream.

3. a kind of efficient random physical layer key generation method based on vector quantization according to claim 2, is characterized in that, described step 1 comprises the steps:

Step 1.1, input: the RSS array X={x collecting ₁, x ₂..., x _k, x _i∈ Z;

Step 1.2, piecemeal: RSS value is carried out to piecemeal, piece is called to block, k RSS value will belong in different block, and the length that makes each block is b,, suppose total t block;

Step 1.3, use label function R (x _i) RSS value is quantized, the RSS value not being quantized is abandoned;

Step 1.4, definition m are improvement factor, wireless communication Party A according to label function generate 0,1 sequence, run into continuous m position identical 0 or 1, the sequence number of interposition is recorded in to L _{a_to_b}in array, generate L _{a_to_b}={ l ₁, l ₂..., l _a, and send to wireless communication Party B;

Step 1.5, wireless communication Party B generate 0,1 sequence according to label function, run into continuous m position identical 0 or 1, the sequence number of interposition is recorded in to L _{b_to_a}in array, and and L _{a_to_b}array contrasts rejects inconsistent sequence number, generates new L _{b_to_a}array also sends to wireless communication Party A.

Label function R (x in described step 1.3 _i) be:

Definition with reference to boundary array is $Q+=\{q_1^{+},q_2^{+},...,q_t^{+}\}$ With $Q-=\{q_1^{-},q_2^{-},...,q_t^{-}\},$ Wherein, $q_i^{+}={\mathrm{mean}}_i+\mathrm{\&alpha;}*\mathrm{std}\_{\mathrm{derivation}}_i,q_i^{-}={\mathrm{mean}}_i-\mathrm{\&alpha;}*\mathrm{std}\_{\mathrm{derivation}}_i,$ Mean _ibe the RSS mean value of i block, α is the fluctuation factor and α ∈ (0,1), std_derivation _iit is the RSS standard deviation of i block.

4. a kind of efficient random physical layer key generation method based on vector quantization according to claim 2, is characterized in that, described step 2 comprises the steps:

Effective RSS array Y after inconsistency is removed of step 2.1, input and will carry out the dimension N of vector quantization;

Step 2.2, for the each y in array Y _iall set up N n dimensional vector n;

Step 2.3, to each y _iall apply N Dimension Vector Quantization of Linear Prediction device Q _n(y _i), 0,1 bit stream producing after quantizing is as output.

RSS array Y in described step 2.1 is: Y={y ₁, y ₂..., y _d, y _i∈ Z, the RSS value in Y array is L _{b_to_a}the corresponding RSS value of each sequence number in array.

In described step 2.2, be y _ithe N n dimensional vector n of setting up is:

$<y_i,y_{(i+{\mathrm{\&Delta;}}_1)\mathrm{mod}d},y_{(i+{\mathrm{\&Delta;}}_1+{\mathrm{\&Delta;}}_2)\mathrm{mod}d},...,y_{(i+{\mathrm{\&Delta;}}_1+...+{\mathrm{\&Delta;}}_{N-1})\mathrm{mod}d}>$

Wherein, Δ={ Δ ₁, Δ ₂..., Δ _n-1the component interval of N component in N n dimensional vector n, Δ _jit is the sequence number interval between j component and j+1 component.

N Dimension Vector Quantization of Linear Prediction device Q in described step 2.3 _n(y _i) be: $Q_N\left(y_i\right)=R\left(y_i\right)R\left(y_{(i+{\mathrm{\&Delta;}}_1)\mathrm{mod}d}\right)...R\left(y_{(i+{\mathrm{\&Delta;}}_1+...+{\mathrm{\&Delta;}}_{N-1})\mathrm{mod}d}\right),$

Wherein, y _ifor the RSS value of N Dimension Vector Quantization of Linear Prediction device input.

5. a kind of efficient random physical layer key generation method based on vector quantization according to claim 2, it is characterized in that, Fuzzy extractor in described step 3 can be defined as: <Gen, Rep>, parameter is (M, l, t), wherein, process Gen is probability generative process, and process Rep is certainty recovery process, and M is incoming bit stream set, l is incoming bit stream length, and t is the ultimate range restriction between wireless communication Party A and the message of wireless communication Party B input.

6. a kind of efficient random physical layer key generation method based on vector quantization according to claim 5, it is characterized in that, first described process Gen is operated by wireless communication Party A, then the parameter of output is passed to wireless communication Party B, wireless communication Party B is again according to process Rep operation, and wireless communication Party A and wireless communication Party B produce on all four random secret information R as encryption key.

7. a kind of efficient random physical layer key generation method based on vector quantization according to claim 5, is characterized in that, described process Gen is: (R, P) ← Gen (w ₀), wherein, input parameter w ₀for the bit stream that length is l, output parameter R is that length is the random secret information of l, and output parameter P is public information.

8. a kind of efficient random physical layer key generation method based on vector quantization according to claim 5, is characterized in that, described process Rep is: R ← Rep (w', P), wherein, input parameter w' is that length is l and and w ₀distance be not more than the bit stream of t, input parameter P is the public information for error correction that Gen process generates, output parameter R is that length is the random secret information identical with Gen process of l.

9. a kind of efficient random physical layer key generation method based on vector quantization according to claim 7, is characterized in that, described process Gen can preferentially adopt and realizes with the following method but be not limited to following method: utilize Bose-Chaudhuri Hocquenghem error correction codes to w ₀just error correction, the error correcting code of generation, as P, utilizes SHA-1 algorithm to w ₀carry out Hash processing, the input of Hash is 1:1 with output Length Ratio, generates the strong key of randomness as R.

10. a kind of efficient random physical layer key generation method based on vector quantization according to claim 8, it is characterized in that, described process Rep can preferentially adopt and realizes with the following method but be not limited to following method: utilize the error correction sequence P producing in Bose-Chaudhuri Hocquenghem error correction codes and Gen process to carry out error correction to w', recover w ₀sequence, utilizes SHA-1 algorithm to w ₀carry out Hash processing, the input of Hash is 1:1 with output Length Ratio, generates the strong key of randomness as R.

Images