Digital signature method and apparatus

Abstract

A method for signing and subsequently verifying a digital message, including the following steps: generating an irreducible monic polynomial f(x) of degree n in a ring Fq[x]; generating an irreducible monic polynomial F(y) of degree n in a ring Fq[y]; producing first and second finite fields as Fq[x]/(f(x)) and Fq[y]/(F(y)), respectively; producing a secret isomorphism from the first finite field to the second finite field; producing and publishing a public key that depends on F(y); producing a private key that depends on the secret isomorphism; producing a message digest by applying a hash function to the digital message and the public key; producing a digital signature using the message digest and the private key; and performing a verification procedure utilizing the digital signature and the public key.

Description

FIELD OF THE INVENTION

This invention relates to the field of cryptography and, more particularly, to a public key digital signature technique and system.

BACKGROUND OF THE INVENTION

Public key digital signatures are important for secure exchange of information between plural parties, for example between computers or mobile devices, between a smart card and a terminal, etc.

An earlier digital signature and authentication method and apparatus was described in U.S. Pat. No. 7,308,097, assigned to the same assignee as the present Application. Reference can also be made to “NTRUSign: Digital Signatures Using the NTRU Lattice”, J. Hoffstein, N. Howgrave Graham, J. Pipher, J. Silverman, and W. Whyte, Topics In Cryptology-CT-RSA 2003, Lecture Notes in Computer Science, Vol. 2612, Springer, Berlin, 2003.

The signing technique in the '097 Patent uses a mixing system based on multiplication in a ring and reduction modulo an ideal q in that ring; while the verification technique uses special properties of products of elements whose validity depends on elementary probability theory. The security of the identification/digital signature scheme comes from the interaction of reduction modulo q and the difficulty of forming products with special properties. In an embodiment of the digital signature scheme of the '097 Patent, the security also relies on the experimentally observed fact that for most lattices, it is very difficult to find a vector whose length is only a little bit longer than the shortest vector, and it is also difficult to find a lattice vector that is quite close to a randomly chosen nonlattice vector.

An improvement over the technique of the '092 Patent, which had reduced complexity and computational requirements for key generation and signing, was disclosed in copending U.S. patent application Ser. No. 14/544,426, assigned to the same assignee as the present Application, published as U.S. Patent Application Publication No. US2015/0229478, incorporated herein by reference. In a form of that invention, sometimes referred to a “pqNTRUSign” (mark of Security Innovation, Inc.), a method is set forth for signing and subsequently verifying a digital message, comprising the following steps implemented using at least one processor-based subsystem: selecting parameters including an integer q and a relatively smaller integer p that is coprime with q; generating random polynomial f relating to p and random polynomial g relating to q; producing a public key that includes h, where h is equal to a product that can be derived using g and the inverse off mod q; producing a private key from which f and g can be derived; storing the private key and publishing the public key; producing a message digest by applying a hash function to the digital message; producing a digital signature using the message digest and the private key; and performing a verification procedure utilizing the digital signature and the public key to determine whether the signature is valid.

It is among the objectives hereof to devise a digital signature method and system that has advantages over existing digital signature techniques, including those of the type described hereinabove.

SUMMARY OF THE INVENTION

The present invention utilizes, inter alia, a secret isomorphism between two finite fields to create a signature scheme.

In accordance with an embodiment of the invention, a method is set forth for signing and subsequently verifying a digital message, comprising the following steps implemented using at least one processor-based subsystem: generating an irreducible monic polynomial f(x) of degree n in a ring F_q[x]; generating an irreducible monic polynomial F(y) of degree n in a ring F_q[y]; producing first and second finite fields as F_q[x]/(f(x)) and F_q[y]/(F(y)), respectively; producing a secret isomorphism from the first finite field to the second finite field; producing and publishing a public key that depends on F(y); producing a private key that depends on said secret isomorphism; producing a message digest by applying a hash function to the digital message and the public key; producing a digital signature using the message digest and the private key; and performing a verification procedure utilizing the digital signature and the public key to determine whether the signature is valid.

In a form of the embodiment just set forth, the first and second finite fields are designated, respectively, as x-space and y-space, and a further step comprises generating a specified lattice in x-space and using said isomorphism to generate a corresponding lattice in y-space, and the step of producing said digital signature includes initially producing a signature in the x-space lattice and then producing the digital signature in the corresponding y-space lattice.

In accordance with another embodiment of the invention, a method is set forth signing and sending a digital message, including the following steps implemented using at least one processor-based subsystem: generating an irreducible monic polynomial f(x) of degree n in a ring F_q[x]; generating an irreducible monic polynomial F(y) of degree n in a ring F_q[y]; producing first and second finite fields as F_q[x]/(f(x)) and F_q[y]/(F(y)), respectively; producing a secret isomorphism from the first finite field to the second finite field; producing and publishing a public key that depends on F(y); producing a private key that depends on said secret isomorphism; producing a message digest by applying a hash function to the digital message and the public key; producing a digital signature using the message digest and the private key; and transmitting the digital signature.

In accordance with a further form of the invention, a system is set forth for signing and subsequently verifying a digital message, including: at least one processor-based subsystem that is programmed with instructions that cause the at least one processor subsystem to implement the following steps: generating an irreducible monic polynomial f(x) of degree n in a ring F_q[x]; generating an irreducible monic polynomial F(y) of degree n in a ring F_q[y]; producing first and second finite fields as F_q[x]/(f(x)) and F_q[y]/(F(y)), respectively; producing a secret isomorphism from the first finite field to the second finite field; producing and publishing a public key that depends on F(y); producing a private key that depends on said secret isomorphism; producing a message digest by applying a hash function to the digital message and the public key; producing a digital signature using the message digest and the private key; and performing a verification procedure utilizing the digital signature and the public key to determine whether the signature is valid.

As described herein (see e.g. Appendix I), for reasons including the non-linear nature of the homomorphic encryption map, the effectiveness of attacks on the signature scheme of the invention is lessened, which enhances security and allows for smaller signatures and improved operating characteristics compared to existing signature schemes.

Further features and advantages of the invention will become more readily apparent from the following detailed description when taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of a system that can be used in practicing embodiments of the invention.

FIG. 2 shows an example of a first finite field (also sometimes designated as x-space) and a second finite field (also sometimes designated as y-space), that is useful in understanding operation of the invention.

FIG. 3 again illustrates the finite fields of the FIG. 2 example, and also illustrates an exemplary isomorphism between polynomials of the respective finite fields, and shows some of the mappings between polynomials of the respective finite fields for an example in which the isomorphism is defined by a particular function that maps from polynomials of the first finite field (x-space) to polynomials of the second finite field (y-space).

FIG. 4A shows, on the left, a table of polynomial values in x-space, evaluated at an arbitrary value of x and, on the right, a table of corresponding (that is, mapped in accordance with the exemplary isomorphism) polynomial values in y-space. The tables illustrate the scrambling of the values that occurs in y-space.

FIG. 4B shows tables similar to those of FIG. 4A, but for a different group of polynomials. Again, the tables illustrate the scrambling of values that occurs in y-space.

FIG. 5 is a flow diagram of a public key digital signature technique which, when taken with the subsidiary flow diagrams referred to therein, can be used in implementing embodiments of the invention.

FIG. 6 is a flow diagram, in accordance with an embodiment hereof, of a routine for key generation.

FIG. 7 is a flow diagram, in accordance with an embodiment hereof, of a routine for signing a digital message.

FIG. 8 is a flow diagram, in accordance with an embodiment hereof, of a routine for verification of a digital signature.

FIG. 9, which includes FIGS. 9A and 9B placed one below another, is a flow diagram, in accordance with an embodiment hereof, of a further, and more detailed, routine for signing and subsequently verifying a digital message.

DETAILED DESCRIPTION

FIG. 1 is a block diagram of an example of a system that can be used in practicing embodiments of the invention. Two processor-based subsystems 105 and 155 are shown as being in communication over a channel 50, which may be, for example, any wired or wireless communication channel such as a telephone or internet communication channel in, for example, a cloud-based system. In the example hereof, the channel can be considered a secure or an insecure channel. The subsystem 105 includes processor 110 and the subsystem 155 includes processor 160. The subsystems can typically comprise mobile devices, computers, or terminals. When programmed in the manner to be described, the processors 110 and 160 and their associated circuits can be used to implement an embodiment of the invention and to practice an embodiment of the method of the invention. The processors 110 and 160 may each be any suitable processor, for example an electronic digital processor or microprocessor. It will be understood that any general purpose or special purpose processor, or other machine or circuitry that can perform the functions described herein, electronically, optically, or by other means, can be utilized. The subsystem 105 will typically include memories 123, clock and timing circuitry 121, input/output functions 118 and display 125, which may all be of conventional types. Inputs can include a touchscreen/keyboard input as represented at 103. Communication is via transceiver 135, which may comprise any suitable device for communicating signals.

The subsystem 155 in this illustrative embodiment can have a similar configuration to that of subsystem 105. The processor 160 has associated input/output circuitry 164, memories 168, clock and timing circuitry 173, and a display 176. Inputs include a touchscreen/keyboard 155. Communication of subsystem 155 with the outside world is via transceiver 162.

Co-inventors hereof, Jeffrey Hoffstein and Joseph Silverman, are inventors of copending PCT International Application No. PCT/US2016/041598, published as PCT Publication No. WO/2017/008043, entitled “Homomorphic Encryption”, and said PCT Publication is incorporated herein by reference. The '598 Application discloses systems, methods, and computer-readable storage devices storing instructions for homomorphic encryption via finite ring isomorphisms that includes constructing an isomorphism and an inverse isomorphism and using these for encryption that is homomorphic, in that certain operations can be performed on encrypted data without the need for first decrypting the data.

Consider f=Z/qZ[x]/(f(x)), a finite field F_qⁿ of order qⁿ. f(x) and F(y) define two copies of f_qⁿ, and Z/qZ[x]/(f(x))≅Z/qZ[y]/(F(y)) is a finite field isomorphism under a secret mapping x→φ(y).

FIG. 2 illustrates an example of the first finite field F_q(x)/f(x) (represented in conjunction with the lefthand box of the Figure) and the second finite field f_q(y)/F(y) (represented in conjunction with the righthand box of the Figure). [In mathematical language, the first finite field is the quotient of the ring of polynomials in x, with coefficients mod q, by the ideal generated by the irreducible polynomial f(x), and the second finite field is the quotient of the ring of polynomials in y, with coefficients mod q, by the ideal generated by the irreducible polynomial F(y).] In this example, n=5 and there are 3125 (equals 5⁵) polynomials listed in each box in an indexed order, with increasing coefficient values. In this example, the polynomial f(x) is x⁵+x⁴+4x³+x²+4x+1, and the polynomial F(y) is y⁵+y⁴+3y³+2y²+2y+4.

In the illustrative example of FIG. 3, which again shows the finite fields, the secret isomorphism between the finite fields is selected to be represented by the polynomial 4(y)=4y⁴+4y²+4y+3; that is, x in the finite field F_q(x)/f(x) (which is designated as x-space) maps to 4y⁴+4y²+4y+3 in the finite field F_q(y)/F(y) (which is designated as y-space). The arrows shown between x-space and y-space in FIG. 3 illustrate some of the mappings for this isomorphism.

As an example, the following calculations show how the index #726 (polynomial 1+0x+4x²+0x³+1x⁴) of the x-space finite field, maps to the index #837 (polynomial 2+2y+3y²+1y³+1y⁴) of the y-space finite field. Using 3+4y+4y²+4y⁴ for x in the expression 1+4x²+x⁴ gives 1+4(3+4y+4y²+4y⁴)²+(3+4y+4y²+4y⁴)⁴. When this is expanded and coefficients are reduced mod 5, the result is the long polynomial:

A(y)=3+3y+y²+4y³+y⁴+4y⁵+4y⁶+y⁷+y⁹+4y¹⁰⁺²y¹¹+4y¹²+4y¹³+4y¹⁴+y¹⁶

and this can be written as:

A(y)=(1+y)²(4+2y+2y²+3y³+y⁴+y⁵)(4+4y+2y²+3y³+2y⁴+4y⁵+2y⁶+2y⁷+2y⁸+y⁹)+2+2y+3y²+y³+y⁵ mod 5.

This expression [A(y)] must be modded by F(y), which, in this example, is y⁵+y⁴+3y³+2y²+2y+4, so:

A(y)=(4+4y+2y²+3y³+2y⁴+4y⁵+2y⁶+2y⁷+2y⁸+y⁹)(1+y)²(F(y))+2+2y+3y²+y³+y⁴ mod 5.

So, in other words, when A(y) is divided by F(y) mod 5, the remainder is 2+2y+3y²+y³+y⁴, which, in the example of FIG. 3, is the polynomial index #837 in y-space. Ergo, polynomial index #726 in x-space maps to polynomial index #837 in y-space.

The foregoing also demonstrates that 1+4x²+x⁴ maps to 1+4 (image of x)²+(image of x)⁴ which demonstrates the homomorphic mapping property; that is, additive and multiplicative structure is preserved under the mapping.

Also, x maps to image of x=3+4y+4y²+4y⁴, which means that f(x)=x⁵+x⁴+4x³+x²+4x+1 maps to F(y)=(3+4y+4y²+4y⁴)⁵+(3+4y+4y²+4y⁴)⁴+4(3+4y+4y²+4y⁴)³+(3+4y+4y²+4 y⁴)²+4(3+4y+4y²+4y⁴)+1.

Since f(x) is zero in x-space (since F_q[x] is modded by f(x)), it should map to zero in y-space. The following demonstrates that it does.

Expanding F(y)mod 5 gives 4+4y+y³+3y⁴+2y⁵+4y⁶+2y⁷+y⁸+4y⁹+y¹⁰+2y¹¹+4y¹³+y¹⁶+4y²⁰

and if this is factored mod 5, one gets

4(4+2y+2y²+3y³+y⁴+y⁵)(4+2y²+2y³+4y⁴+y⁵)(1+3y+2y³+y⁴+3y⁵+2y⁶+2y⁷+y⁸+y¹⁰).

F(y)=4+2y+2y²+3y³+y⁴+y⁵

So A(y)=4(4+2y²+2y³+4y⁴+y⁵)(1+3y+2y³+y⁴+3y⁵+2y⁶+2y⁷+y⁸+y¹⁰)F(y)mod 5.

Stated another way, when the polynomial that is mapped to [called A(y)] is divided by F(y), the remainder is zero, so zero maps to zero, as it should.

The tables of FIGS. 4A and 4B illustrate the property hereof that the isomnrphism doesn't “see” the sizes of the coefficients of the x polynomials, which is an indication that images in y-space are not revealing of relevant information about x-space. FIGS. 4A and 4B show two examples of the principle. Both examples involve the same finite fields and the same secret mapping isomorphism as in FIGS. 2 and 3.

In FIG. 4A, twenty five polynomials in x-space are defined by the exemplary general polynomial f(x)=a₄x⁴+a₃x³+2x²+2x+2, where a₄ and a₃ can each take on any of the five values 0, 1, 2, 3, or 4, so that there are 25 possible polynomials. In the FIG. 4A table on the left a₄ varies from 0 to 4 on the horizontal axis and a₃ varies from 0 to 4 on the vertical axis. The numerical value in each of the 25 boxes is the value of the polynomial for the particular box, evaluated at x=5. For example, the upper leftmost box of the table, where a₄=0 and a₃=0, accordingly is associated with the polynomial f(x)=2x²+2x+2 which, evaluated at x=5, is (2)(25)+(2)(5)+2=62. As another example, consider the lower rightmost box of the table, where a₄=4 and a₃=4, so the associated polynomial is f(x)=4x⁴+4x³+2x²+2x+2 which, evaluated at x=5, is 4(625)+4(125)+2(25)+2(5)+2=3062.

In the FIG. 4A table on the right, each of the 25 boxes is associated with the polynomial F(y) that is the y-space image of the x-space polynomial f(x) in the corresponding box of the table on the left. The number in each box is the value of the associated polynomial, evaluated at y=5.

In FIG. 4B, twenty five polynomials in x-space are defined by another exemplary polynomial f(x)=2x⁴+2x³+a₂x²+a₁x+2, where a₂ and a₁ can, again, take on any of the five values, 0, 1, 2, 3, or 4 so that there are 25 possible polynomials. In the FIG. 4B table on the left a₂ varies from 0 to 4 on the horizontal axis and a₁ varies from 0 to 4 on the vertical axis. As before, the numerical value in each of the 25 boxes is the value of the polynomial for the particular box, evaluated at x=5. In this case, for example, the upper leftmost box of the table, where a₂=0 and a₁=0, accordingly is associated with the polynomial f(x)=2x⁴+2x³+2 which, evaluated at x=5, is (2)(625)+(2)(125)+2=1502. As another example, consider the lower rightmost box of the table, where a₂=4 and a₁=4, so the associated polynomial is f(x)=2x⁴+2x³+4x²+4x+2 which, evaluated at x=5, is 2(625)+2(125)+4(25)+4(5)+2=1622.

In the FIG. 4B table on the right, each of the 25 boxes is again associated with the polynomial F(y) that is the y-space image of the x-space polynomial f(x) in the corresponding box of the table on the left. The number in each box is the value of the associated polynomial, evaluated at y=5.

Accordingly, FIGS. 4A and 4B illustrate the scrambling that occurs in y-space as a result of the defined process using the secret isomorphism. As seen in the examples of these Figures, although the values of the selected polynomials in x-space are generally ordered in their tables, the y-space polynomials have no discernable order in their respective tables.

FIG. 5 illustrates a basic procedure that can be utilized with a public key digital signature technique, and refers to routines illustrated by other referenced flow diagrams which describe features in accordance with an embodiment of the invention. Reference can also be made to Appendix I for further details of embodiments of the invention. The block 510 represents the generating of the public key and private key signals and data, and the publishing of the public key. The routine of an embodiment thereof is described in conjunction with the flow diagram of FIG. 6. In the present example, this operation can be performed, for example, at the processor-based subsystem 105 of FIG. 1. The public key information can be published; that is, made available to any member of the public or to any desired group to whom the private key holder desires to send the digital signatures. Typically, although not necessarily, the public key may be made available at a central public key library facility or website where a directory of public key holders and their public keys are maintained.

The block 550 represents a routine that can be employed (that is, in this example, by the user of processor-based subsystem 105 of FIG. 1) for signing the digital message. This routine, in accordance with an embodiment of the invention, is described in conjunction with the flow diagram of FIG. 7. In this example, the digital signature is then transmitted over the channel 50 (FIG. 1), although it will be understood that the digital signature can, if desired, be stored and subsequently retrieved and verified, depending on the application or task being undertaken.

The block 570 represents a routine that can be employed (that is, in this example, by the user of processor-based subsystem 155 of FIG. 1) for using, inter alia, the public key to implement a verification procedure to either accept or reject the digital signature. This routine, in accordance with an embodiment of the invention, is described in conjunction with the flow diagram of FIG. 8.

FIG. 6 is a flow diagram of a routine, represented generally by the block 510 of FIG. 5, in accordance with an embodiment of the invention, for implementing key generation. Reference can also be made to Appendix I. The block 605 represents the inputting of initial parameters as, for example, described in Appendix I. The block 610 represents generating of an irreducible monic polynomial f(x) of degree n in a ring F_q(x), and the block 620 represents generating of an irreducible monic polynomial F(y) of degree n in a ring F_q(y). Then, as represented by the block 630, first and second finite fields are produced as F_q(x)/f(x) and F_q(y)/F(y), respectively. Next, as represented by the block 640, an isomorphism is produced from the first finite field to the second finite field. This can be, for example, an isomorphism e.g. in the form of a matrix or a polynomial of the type illustrated in conjunction with FIG. 3 as the mapping polynomial φ, but of a magnitude commensurate with a desired degree of security. A public key that depends on F(y) can then be output (block 650) and, typically, published. A private key, which depends on the isomorphism, is stored (block 660) for subsequent use.

FIG. 7 is a flow diagram of a routine, represented by the block 550 of FIG. 5, in accordance with an embodiment of the invention, for implementing the signing of a digital message that is to be stored and/or transmitted. Reference can also be made to Appendix I. The block 750 represents the inputting of the digital message that is to be encoded. Then, as represented by the block 760, a message digest is produced by applying a hash function to the digital message and the public key. The encoded digital signature is then produced (block 770) using the message digest and the private key. The encoded digital signature can then be output (block 780). If desired, an optional rejection sampling technique, represented by loop 775, can be employed, as set forth in copending U.S. patent application Ser. No. 14/121,041 (published as U.S. Patent Application Publication No. US2015/0033025) to check a candidate signature to determine if it may leak information that could be used in a transcript of signatures to reveal information about the public key and, if so, modifying one or more parameters used in producing the candidate digital signature, the procedure being repeated until an acceptable signature is obtained.

FIG. 8 is a flow diagram of a routine, represented by the block 570 of FIG. 5, in accordance with an embodiment of the invention, for implementing verification to accept or reject the digital signature. The block 810 represents the receiving of the digital signature, and the block 820 represents the inputting of the public key. Then, the decision block 830 represents the determination of whether the collective components of these and other inputs satisfy predetermined size and lattice presence conditions. If not, the signature is rejected. If so, however determination is also made (decision block 840) as to whether a predetermined congruence condition is met. If so, the digital signature is considered verified.

FIG. 9 is a flow diagram of a routine in accordance with a further embodiment of the invention for encoding and decoding a digital message. Reference can be made to Appendix I for further details. The block 910 represents inputting parameters such as p, q, n, and other parameters as described in Appendix I. The finite fields F_q(x)/(f(x)) (x-space, as described above) and F_q(y)/(F(y)) (y-space, as described above) are produced, as represented by the block 920. Next, the block 930 represents the selection of secret a(x), b(x) in x-space with small coefficients, and the computation of a secret h(x)=(pa(x))⁻¹b(x). (As used herein in this context, a small coefficient is an integer with absolute value less than c, where c is significantly less than q, an example being q=2048, c=4.) Then, as represented by the block 940, a secret isomorphism (such as φ in FIG. 3) is produced, mapping h(x) (in x-space) to H(y) (in y-space). A public key that includes H(y) can then be published (block 950). Then, as represented by the block 955, a lattice L_h is produced in x-space and a lattice L_H is produced in y-space. L_h is the set of all pairs of polynomials (s(x), t(x)) of degree less than or equal to n−1 such that t(x) is equivalent to S(x)h(x) mod(f(x),q), and L_H is the set of all pairs of polynomials (S(y), T(y)) of degree less than or equal to n−1 such that T(y) is equivalent to S(y)H(y) mod (F(y),q).

A message digest is obtained from a hash of the digital message and the public key (block 960). In this embodiment, the message digest is a 2n dimensional polynomial (d₀, d₁, . . . d_n-1, e₀, e₁ . . . e_n-1), with coefficients chosen from the interval [−p/2, p/2]. The block 970 represents finding (s(x), t(x)) in L_n such that the 2n dimensional vector of coefficients from (s(x), t(x)) satisfies the correct bound on absolute values, and is congruent to a specified 2n dimensional vector determined by the message digest and the polynomials C_i(y). The coefficients of this linear combination are output as the digital signature. Then, as represented by the block 985, the digital signature is received and subject to a verification procedure. In this embodiment, a determination is made (decision block 995) as to whether the linear combination of C_i(y) is in L_H and, also, whether the 2n dimensional vector of coefficients has absolute values in a predetermined range and is congruent to (d₀, d₁ . . . d_n-1, e₀, e₁, . . . e_n-1)mod p. If both criteria are satisfied, the digital signature is verified. If not, it is rejected.

The invention has been described with reference to particular preferred embodiments, but variations within the spirit and scope of the invention will occur to those skilled in the art. For example, while a digital signature technique has been described, it will be understood that an authentication procedure of the challenge-response-verification type can alternatively be implemented, using the technique hereof and employing the challenge as the message to be signed. Also, it will be understood that coefficients of polynomials can alternatively be represented in other forms including, but not limited to, matrices.

APPENDIX I

There are a number of possible methods of using a secret isomorphism between two finite fields to create a signature scheme. In the following, a method we call “pqFF-Sign” is set forth.

The approach utilizes the existing transcript-secure signature scheme pq-NTRUSign and also utilizes a finite field isomorphism. The pq-NTRUSign signature scheme is employed in a finite field _qⁿ, and then the signature is homomorphically mapped to a different copy of the field _qⁿ. Verification is still possible due to the homomorphic property of the map, but various lattice attacks that may have been possible on pq-NTRUSign are blunted or eliminated due to the non-linear nature of the homomorphic encryption map.

n a dimension parameter.

q a prime (or prime power) greater than a constant times n.

p a small prime (often taken to be 2).

f(x)ε_q[x] a short irreducible monic polynomial of degree n.

F(y)ε_q[y] a random irreducible monic polynomial of degree n.

ψ(x)εF(y)ε_q[y]. The map xφ(y) induces an isomorphism _q[x]/(f(x))→_q[y]/(F(y)).

ψ(x)ε_q[x] the map yψ(x) induces the inverse isomorphism _q[y]/(F(y))→_q[x]/(f(x)).

a(x), b(x)ε_q[x] short irreducible monic polynomials of degree n.

h(x)ε_q[x]≡b(x)·(pa(x))⁻¹ (mod q).

H(y)ε_q[y]≡h(φ(y)) (mod q, F(y)).

U is an n-by-n matrix, small entries, invertible mod q.

We will lift mod q polynomials to polynomials having integer coefficients in the range (−½q,½q]. Define rings and fields

$f_{}=\frac{\mathbb{Z}\left[x\right]}{\left(f\left(x\right)\right)},F_{}=\frac{\mathbb{Z}\left[y\right]}{(F\left(y\right)},{f{,}_q}_{}=\frac{{}_q\left[x\right]}{\left(f\left(x\right)\right)},{F,q}_{}=\frac{{}_q\left[y\right]}{\left(F\left(y\right)\right)}.$

And define lattices

L _h={(u,v)ε_f²:v≡h·u(mod q)},

L _H={(U,V)ε_F²:V=H·U(mod q)}.

We use U to define polynomials c₁(x), . . . , c_n(x)ε_q[x] of degree less than n by

$\left(\begin{array}{c}{c}_1\left(x\right)\\ c_2\left(x\right)\\ ⋮\\ c_n\left(x\right)\end{array}\right)\equiv U^{-1}\left(\begin{array}{c}x\\ x^2\\ ⋮\\ x^n\end{array}\right)\left(\mathrm{mod}q,f\left(x\right)\right).$

For 1≦j≦n we let

C _j(y)=c_j(φ(y))ε_F,q

be the corresponding polynomials in the y-field.

Security Assumption

As U ranges over matrices with small coefficients that are invertible modulo q, the coefficients of U⁻¹ mod q are uniformly distributed.

The polynomials c₁(x), c₂(x), . . . , c_n(x) form a basis for _f,q and C₁(y), C₂(y), . . . , C_n(y) form a basis for _F,q. Each C_j(y) is the image of the corresponding c_j(x) under the isomorphism that sends xφ(y). This same isomorphism preserves the coefficients of linear combinations of the c_j(x), that is,

Σα_jc_j(x)Σα_jC_j(y).

A key property that the scheme is based on is the fact that as

$U\left(\begin{array}{c}{c}_1\left(x\right)\\ c_2\left(x\right)\\ ⋮\\ c_n\left(x\right)\end{array}\right)\equiv \left(\begin{array}{c}x\\ x^2\\ ⋮\\ x^n\end{array}\right)\left(\mathrm{mod}q,f\left(x\right)\right),$

and the coefficients of U are small, that each xⁱ, for 1≦i≦n is expressible as a linear combination of the c_j(x) with small coefficients. From this it follows that any polynomial in x with small coefficients, of the form t(x)=Σ_i=1ⁿt_ixⁱ, with the t_i small, can in turn be written as a polynomial in c_j(x) with small coefficients.

We will find it convenient to use polynomials reduced modulo f(x), which will necessarily have degree less than or equal to n 1, that is, of the form r(x)=Σ_i=0^n-1r_ixⁱ, with the r_i short. Recall that f(x)=xⁿ+f_n-1x^n-1+ . . . +f₁x±1, where f_iε{1, 0, −1}. Consequently

±1≡f₁x− . . . −xⁿ(mod f(x)).

and thus

$r\left(x\right)\equiv \sum _{i=1}^nr_i^{\prime }x^i\equiv \pm r_0\left(-f_1x-\dots -x^n\right)+\sum _{i=1}^{n-1}r_ix^i(\mathrm{mod}\left(f\left(x\right)\right),$

where the r_i′ are also short, is expressible as a short linear combination of the c_j(x).

Importantly, as the product of any two short polynomials in x remains short, such a product will also be writeable as a short linear combination of the c_j(x). The reverse does not hold: with high probability a random linear combination of c_j(x) with small coefficients will not equal a polynomial in x with small coefficients.

It is this property, that a product of short linear combinations of the c_j(x) that correspond to short polynomials in x can again be written as a short linear combination of the c_j(x), that allows the signer to solve a congruential lattice problem in L_h (just as in pq-NTRUSign) and then map the corresponding solution, with the same coefficients, back to L_H.

There are two main security concerns that determine parameters in pq-NTRUSign. One is the problem of recovering the private key from the public NTRU key, and the other is the problem of forgery. Of these, the one that has the biggest impact on parameter size is the public key to private key problem. This is because, to make rejection sampling efficient, the q needs to be chosen large compared to n. This makes the lattice problem somewhat easier and forces an increase in the size of n. The forgery problem requires smaller parameters to achieve the same security levels.

In this context there appear at first to be two NTRU-type problems: Recovering a(x), b(x) from h(x), and recovering the corresponding polynomials A(y), B(y) from H(y).

The h(x) is private, and only revealed if the underlying isomorphism is discovered, in which case the scheme is considered broken. So this lattice problem does not arise.

On the other hand, the H(y) is public, but the corresponding problem of recovering A(y), B(y) from H(y) is not a lattice reduction problem as A(y), B(y) are generic polynomials with coefficients mod q, and not short. And as they are not short, recovery of them would not lead to any advantage.

There is a lattice attack to recover the matrix U from the C_j(y), which would suffice to break the scheme, but the dimension of the lattice required to accomplish this is greater than n².

For this reason, it appears that it will suffice to set parameters to avoid forgery attacks, which should allow for smaller signatures and better operating characteristics.

Private Information: f(x), a(x), b(x), h(x), c₁(x), . . . , c_n(x) and U.

Public Information: F(y), H(y) and C₁(y), . . . , C_n(y).

Digital Document Hash: A pair of mod p vectors δ, ε obtained by applying a hash function to the document being signed:

δ=δ₁, . . . ,δ_n)ε(−½p,½p]ⁿ,

ε=ε₁, . . . ,ε_nε(−½p,½p]ⁿ,

Signature: A pair of vectors

δ=(δ₁, . . . ,δ_n)ε(−½q,½q]ⁿ,

ε=(ε₁, . . . ,ε_n)ε(−½q,½q]ⁿ,

Verification: A signature on the document hash (δ,ε) is valid if it satisfies the following three conditions:

• • (1) δ_i≡δ_i (mod p) and ε_i≡ε_i (mod p) for all 1≦i≦n.
  • (2) |δ_i|≦½q−B and |ε_i|≦½q−B for all 1≦i≦n.
  • (3) Let

S(y)=δ₁C₁(y)+ . . . +δ_nC_n(y),

and

T(y)=ε₁C₁(y)+ . . . +ε_nC_n(y).

Then (S,T)εL_H, i.e.,

T(y)=S(y)H(y)(mod q,F(y)).

Here B is a fixed small integer used to enable rejection sampling.

Signatures are created as in pq-NTRUSign working in the ring _f=[x]/(f(x)), with one change. Rather than creating polynomials with small coefficients relative to the standard basis 1, x, . . . , x^n-1, we instead create polynomials with small coefficients relative to the basis c₁(x), . . . , c_n(x).

Step 1: Choose δ_j at random mod q such that q/2<δ_j≦q/2 and δ_j≡δ_j (mod p) and set

$s_0\left(x\right)=\sum _{j=1}^n{\delta }_jc_j\left(x\right).$

Step 2: Define t₀(x) by

t ₀(x)≡s₀(x)h(x)(mod q)

and write

$t_0\left(x\right)=\sum _{i=0}^{n-1}t_ix^i.$

Then (s₀(x),t₀(x))εL_h.

Step 3: Write

$t_0\left(x\right)=\sum _{j=1}^n{\eta }_jc_j\left(x\right)$

for some η₁, . . . , η_n.

To accomplish this, as described previously, write

$t_0\left(x\right)=\sum _{i=0}^{n-1}t_ix^i\equiv \sum _{i=1}^nt_i^{\prime }x^i\left(\mathrm{mod}f\left(x\right)\right)$

as described previously, and set

(η₁, . . . ,η_n)≡(t′₁, . . . ,t′_n)U(q),

and select representatives for the η_j in the interval (−q/2/q/2]. The η_j will appear to be randomly and uniformly distributed mod q.

Step 4: Construct (u(x), v(x))εL_h such that

$u\left(x\right)=\sum _{j=1}^n{\delta }_j^{\left(u\right)}c_j\left(x\right)\mathrm{and}\upsilon \left(x\right)=\sum _{j=1}^n{\delta }_j^{\left(\upsilon \right)}c_j\left(x\right),$

with δ_j^(u), δ_j^(v) small, δ_j^(u)≡0 (mod p), and δ_j^(v)+ε_j(mod p) for all j.

To construct the desired (u(x), v(x)), search for an appropriate r(x) which is short, and set

u(x)=pr(x)a(x) and v(x)=r(x)v(x).

Such an r(x) must satisfy

$r\left(x\right)b\left(x\right)=\sum _{j=1}^n{\delta }_j^{\left(\upsilon \right)}c_j\left(x\right),$

with the δ_j^(v) small and δ_j^(v)+η_j=ε_j(mod p),

and also satisfy

$\mathrm{pr}\left(x\right)a\left(x\right)=\sum _{j=1}^n{\delta }_j^{\left(u\right)}c_j\left(x\right),$

with the δ_j^(u) small and δ_j^(u)≡(mod p).

As r(x), a(x) are short, r(x)a(x) is also short, and we may write

$r\left(x\right)a\left(x\right)=\sum _{i=0}^{n-1}d_ix_i\in {}_f,$

with the d_i small. Then the δ_j^(u) of

$\mathrm{pr}\left(x\right)a\left(x\right)=\sum _{j=1}^n{\delta }_j^{\left(u\right)}c_j\left(x\right),$

are given by

(δ₁^(u), . . . ,δ_n^(u))=p(d₀, . . . ,d_n-1)U.

As all the d_i and entries of U are small there is no wraparound mod q and each δ_j^(u)≡0 (mod p). Thus for whatever short r(x) we find, the (δ_j^(u)=0 (mod p) condition will hold.

We turn now to finding r(x) short, such that

$r\left(x\right)b\left(x\right)=\sum _{j=1}^n{\delta }_j^{\left(v\right)}c_j\left(x\right),$

with δ_j^(v) short and δ_j^(v)≡ε_j−η_j(mod p).

To accomplish this, write b(x)=Σ_i^n-1b_ixⁱ, set

(b_0,0,b_0,1, . . . ,b_0,n-1)=(b₀,b₁, . . . ,b_n-1),

and define (b_i,0, b_i,1, . . . , b_i,n-1) by

x ⁱ b(x)=b_i,0+b_i,1x+ . . . +b_i,n-1x^n-1.

Let B denote the matrix whose i, j entry is b_i,j, and let

β=BU.

Note that the entries β_i,j of β are small because the b_i,j and the entries of U are small.

For any

$r\left(x\right)=\sum _{i=0}^{n-1}r_ix^i\equiv \sum _{i=1}^nr_i^{\prime }x^i(\mathrm{mod}\left(f\left(x\right)\right)$

we have

$r\left(x\right)b\left(x\right)=\left(r_1^{\prime },r_2^{\prime },\dots ,r_n^{\prime }\right)\beta \left(\begin{array}{c}{c}_1\left(x\right)\\ c_2\left(x\right)\\ ⋮\\ c_n\left(x\right)\end{array}\right).$

To solve for r(x), first define

(r′₁,r′₂, . . . ,r′_n)≡(δ₁^(v),δ₁^(v), . . . ,δ₁^(v))β⁻¹(mod p)

and lift each r′_j to r′_jε(−p/2,p/2].

Then define δ_j9^(v) by

(δ₁^(v), . . . ,δ_n^(v))≡(r′₁, . . . ,r′_n)β.

This accomplishes the goal

$r\left(x\right)b\left(x\right)=\sum _{j=1}^n{\delta }_j^{\left(v\right)}c_j\left(x\right),$

with δ_j^(v)=δ_j^(v) (mod p).

After accomplishing Step 4, we have found a short pair of vectors (u(x), v(x))εL_h with the appropriate congruential properties.

Having done so, set

s(x)=s₀(x)+u(x) and t(x)=t₀(x)+v(x).

Then

$s\left(x\right)=\sum _{j=1}^n\left({\delta }_j+{\delta }_j^{\left(u\right)}\right)c_j\left(x\right)\mathrm{and}t\left(x\right)=\sum _{j=1}^n\left({\eta }_j+{\delta }_j^{\left(v\right)}\right)c_j\left(x\right).$

By our construction, δ_j+δ_j^{(u) and η}_j+δ_j^(v). satisfy the required congruences mod p.

If, in addition, for an appropriate choice of , |δ_j+δ_j^(u)|<q/2−, and |η_j+δ_j^(v)|<q/2−, then we accept the signature and release it. If not, we repeat the process.

An argument very similar to that in pq-NTRUSign shows that this guarantees an information free transcript.

Why does short times short=pretty short?:

We will investigate the size of the coefficients of the remainder when a polynomial b(x) is divided by some other polynomial f(x), and in particular, how the coefficients of the remainder depend on the magnitude of the roots of f(x).

Example

Take n=150 and choose f(x) randomly to have the form

f(x)=x¹⁵⁰+(random trinary polynomial of degree 90).

There are roughly 2¹⁴² such f. We next choose random trinary polynoimals g₁(x), g₂(x) of degree 150 and compute g₁(x)g₂(x) mod f(x). A sequence of random trials of f, g₁, g₂ produced polynomials whose (maximum coefficient, minimum coefficient) were

(343,−484),(255,−235),(607,−486),(411,−441),(552,−560), . . .

The spread of the coefficient range is very dependent on the size of the largest complex root of f(x). These roots will in general be considerably smaller if there is a large gap between the leading coefficient of highest degree (150 in the Example) and the non-zero coefficient of highest degree below the leading coefficient (90 in the Example).

Fix integers m≧n>0. Fix a polynomial

$f\left(x\right)=\prod _{i=1}^n\left(x-{\theta }_i\right)\in ℂ\left[x\right].$

Let

$b\left(x\right)=\sum _{i=0}^{m-1}b_ix^i$

be chosen with each b_i satisfying some probability distribution. Different coefficients may have different distributions, but we assume that they are independent and have mean 0. (In practice, our b(x) will be a product of plaintexts, so it will be a product of t polynomials whose coefficients are independent and more-or-less uniform in some interval. This means that the coefficients of b(x) each satisfy some sort of t-fold hypergeometric distribution, but note that the middle coefficients will be much larger than the ones near the top and the bottom. That is why we allow the coefficients of our b to have different distributions.) The independence means that

E(b_ib_j)=E(b_i)E(b_j)=0 if iδj,

while the numbers E(b_i²) depend on the distributions satisfied by the various b_i.

We perform division with remainder,

b(x)=f(x)q(x)+r(x) with 0≦deg r<n.

As usual, we view the polynomials as vectors,

b=(b₀, . . . ,b_m) and r=(b₀, . . . ,b_n)

We let V denote the vanderMonde matrix of the θ_i's,

$V={\left({\theta }_i^j\right)}_{1\le i\le n0\le j<n}=\left(\begin{array}{cccc}1& {\theta }_1& \dots & {\theta }_1^{n-1}\\ 1& {\theta }_2& \dots & {\theta }_2^{n-1}\\ ⋮& & \ddots & ⋮\\ 1& {\theta }_n& \dots & {\theta }_n^{n-1}\end{array}\right),$

and we set

${\theta }^{\left(j\right)}=\left(\begin{array}{c}{\theta }_1^j\\ {\theta }_2^j\\ ⋮\\ {\theta }_n^j\end{array}\right).$

Then we set

$b\left(\theta \right)=\left(\begin{array}{c}b\left({\theta }_1\right)\\ b\left({\theta }_2\right)\\ ⋮\\ b\left({\theta }_n\right)\end{array}\right)=\sum _{j=0}^{m-1}b_j{\theta }^{\left(j\right)},$

and similarly for r(θ).

We take the relation b(x)=f(x)q(x)+r(x) and substitute x=θ₁, . . . , θ_n. Since f(θ_i)=0, this gives

r(θ_i)−b(θ_i) for all 1≦i≦n.

With our earlier notation, this is simply the equality of vectors

r(θ)=b(θ).

Now we observe that since r has degree at most n−1, we can write r(θ) as

$r\left(\theta \right)=\sum _{j=0}^{n-1}r_j{\theta }^{\left(j\right)}=\mathrm{Vr}.$

Hence

r=V ⁻¹ b(θ).

We now compute the expected value of ∥r∥² as b(x) varies.

$\begin{array}{cc}\begin{array}{c}E\left({r}^2\right)=E\left({V^{-1}b\left(\theta \right)}^2\right)\\ =E{(}^t{b\left(\theta \right)}^tV^{-1}V^{-1}b\left(\theta \right))\\ =E\left(\sum _{j,k=0}^{m-1}b_k^l{\theta }^{\left(k\right)t}V^{-1}V^{-1}b_j{\theta }^{\left(j\right)}\right)\\ =\sum _{j,k=0}^{m-1}{E\left(b_kb_j\right)}^t{\theta }^{\left(k\right)t}V^{-1}V^{-1}{\theta }^{\left(j\right)}\\ =\sum _{j=0}^{m-1}{E\left(b_j^2\right)}^t{\theta }^{\left(j\right)t}V^{-1}V^{-1}{\theta }^{\left(j\right)}\\ =\sum _{j=0}^{m-1}E\left(b_j^2\right){V^{-1}{\theta }^{\left(j\right)}}^2.\end{array}& \left(1\right)\end{array}$

This last formula explains what's going on. If we assume that f(x) is fixed and that deg b(x) is large compared to n=deg f(x), then we obtain the rough, but useful, estimate

$E\left({r}^2\right)\underset{0\le j<m}{\max }\left(E\left(b_j^2\right)·\underset{1\le i\le n}{\max }{{\theta }_i}^j\right).$

Which term dominates will depend on the relative size of E(b_j²) and max |θ_i|^j for 0≦j<m. In our scenario, we have b(x)=a_i(x) . . . a_t(x) with deg a_i≈n, so m≈nt. The coefficients of the a_i are uniform and small, so most of the coefficients of b are roughly C^t. Then E(∥r∥²) is roughly C^t max |θ_i|^nt. So in order for decryption to work, we need roughly

g>(C max|θ_i|ⁿ)^t.

As expected, we get exponential growth in t. But this shows very clearly how the largest root of f(x) has a major influence on the required size of q. Definintion:

Let f(x)εC[x] be a monic polynomial and let θ₁, . . . , θ_n be the roots of f. We let

$\left(f\right)=\underset{1\le i\le n}{\max }{\theta }_i.$

Experiments clearly reveal the effect of the size of the roots of f(x). We fixed an f(x) of degree 11, chose 100 polynomials g(x) of degree 32 with random coefficients in [−2,2] and computed the largest coefficients of g(x) modulo f(x). We used the polynomials

f ₁(x)=x¹¹−x¹⁰+x⁹+x⁶−x⁵+x²−x−1

f ₂(x)=x¹¹+x¹⁰+x⁵−x⁴+x³−x²−x−1

f ₃(x)=x¹¹−x¹⁰+x⁷+x⁶+x⁵−x³−x²−1.

Then

	ƒ	(ƒ)	Avg |g mod ƒ|∞	St. Dev. |g mod ƒ|∞
	ƒ1	1.1835	43.420	16.226
	ƒ2	1.3511	352.250	191.452
	ƒ3	1.4307	1167.720	666.196

We now consider if there is an advantage in taking the non-zero coefficients of f(x) to be in the lower degree terms. So we take f(x) to have the form

f(x)=xⁿ+{tilde over (f)}(x),

where {tilde over (f)}(x) is random trinary of small degree. Simple estimates make it clear that such polynomials tend to have smaller roots than polynomials whose non-zero monomials have higher degree. In order to compare with our experiments, we took polynomials f(x) of degree 11 with non-zero coefficients only on monomials of degree at most 4, more precisely, we took

f(x)=x¹¹+a₄x⁴+a₃x³+a₂x²+a₁x−1

with the a_i randomly chosen from {±1}. The polynomial

f ₄(x)=x¹¹−x⁴+x³−x²++x−1

has

(f₄)=1.18225,

so (f₄) is comparable to (f₁) for the f₁(x) above.

For f₄ and 100 samples, we found

Avg |g mod f₄|_∞=28.450 and St.Dev. |g mod f₄|_∞=15.658.

These may be compared with the roughly similar values 43.4 and 16.2 for f₁. A likely reason for the difference is due to secondary effects due to the other roots. Thus the magnitudes of the roots of f₁ are

1.18,1.18,1.15,1.15,1.08,1.08,1.00,1.00,0.890,0.890,0.578,

while the magnitudes of the roots of f₄ are

1.18,1.18,1.00,1.00,1.00,1.00,1.00,0.953,0.953,0.888,0.888.

So the second largest root of f₁ is significantly larger than the second largest root of f₄.

As the above formula makes clear, the size of the inverse of the vanderMonde matrix V_f also has an effect. We list the sup norm and the spectral radius of V_f⁻¹ for our two example polynomials.

	ƒ1	ƒ4
	Spectral Radius of Vƒ−1	7.766	5.522
	Sup Norm of Vƒ−1	0.666	0.263

We note that the remainder coefficients for division by f₁ and f₄ resemble one another much more closely than do the remainder coefficients for division by f₂ or f₄. This suggests that it is not so much the distribution of non-zero monomials that affects the remainder coefficients as it is the size of the roots of f. However, if one desires to find an f with comparatively small roots, it is definitely advantageous to select f with non-zero monomials only in the lower degree terms.

Claims

1. A method for signing and subsequently verifying a digital message, comprising the following steps implemented using at least one processor-based subsystem: generating an irreducible monic polynomial f(x) of degree n in a ring Fq[x];generating an irreducible monic polynomial F(y) of degree n in a ring Fq[y];producing first and second finite fields as Fq[x]/(f(x)) and Fq[y]/(F(y)), respectively;producing a secret isomorphism from the first finite field to the second finite field;producing and publishing a public key that depends on F(y);producing a private key that depends on said secret isomorphism;producing a message digest by applying a hash function to the digital message and the public key;producing a digital signature using the message digest and the private key; andperforming a verification procedure utilizing the digital signature and the public key to determine whether the signature is valid.

2. The method as defined by claim 1, wherein said first and second finite fields are designated, respectively, as x-space and y-space, and further comprising generating a specified lattice in x-space and using said isomorphism to generate a corresponding lattice in y-space, and wherein said step of producing said digital signature includes initially producing a signature in the x-space lattice and then producing the digital signature in the corresponding y-space lattice.

3. The method as defined by claim 1, wherein said secret isomorphism comprises a polynomial.

4. The method as defined by claim 1, further comprising transmitting the digital signature, and wherein said step of performing a verification procedure includes receiving the transmitted digital signature and performing the verification procedure on the received digital signature.

5. The method as defined by claim 2, further comprising transmitting the digital signature, and wherein said step of performing a verification procedure includes receiving the transmitted digital signature and performing the verification procedure on the received digital signature.

6. The method as defined by claim 1, wherein said digital message comprises a challenge communication from a verifier entity, and wherein said digital signature is transmitted to said verifier entity.

7. The method as defined by claim 2, wherein said digital message comprises a challenge communication from a verifier entity, and wherein said digital signature is transmitted to said verifier entity.

8. A method for signing and sending a digital message, comprising the following steps implemented using at least one processor-based subsystem: generating an irreducible monic polynomial f(x) of degree n in a ring Fq[x];generating an irreducible monic polynomial F(y) of degree n in a ring Fq[y];producing first and second finite fields as Fq[x]/(f(x)) and Fq[y]/(F(y)), respectively;producing a secret isomorphism from the first finite field to the second finite field;producing and publishing a public key that depends on F(y);producing a private key that depends on said secret isomorphism;producing a message digest by applying a hash function to the digital message and the public key;producing a digital signature using the message digest and the private key; andtransmitting the digital signature.

9. The method as defined by claim 8, wherein said first and second finite fields are designated, respectively, as x-space and y-space, and further comprising generating a specified lattice in x-space and using said isomorphism to generate a corresponding lattice in y-space, and wherein said step of producing said digital signature includes initially producing a signature in the x-space lattice and then producing the digital signature in the corresponding y-space lattice.

10. The method as defined by claim 8, wherein said secret isomorphism comprises a polynomial.

11. The method as defined by claim 8, further comprising receiving the digital signature, and employing a further process-based subsystem for performing a verification procedure on the received digital signature.

12. The method as defined by claim 9, further comprising receiving the digital signature, and employing a further process-based subsystem for performing a verification procedure on the received digital signature.

13. A system for signing and subsequently verifying a digital message, comprising: at least one processor-based subsystem that is programmed with instructions that cause the at least one processor subsystem to implement the following steps:generating an irreducible monic polynomial f(x) of degree n in a ring Fq[x];generating an irreducible monic polynomial F(y) of degree n in a ring Fq[y];producing first and second finite fields as Fq[x]/(f(x)) and Fq[y]/(F(y)), respectively;producing a secret isomorphism from the first finite field to the second finite field;producing and publishing a public key that depends on F(y);producing a private key that depends on said secret isomorphism;producing a message digest by applying a hash function to the digital message and the public key;producing a digital signature using the message digest and the private key; andperforming a verification procedure utilizing the digital signature and the public key to determine whether the signature is valid.

14. The system as defined by claim 13, wherein said encoded digital signature is transmitted to a further processor-based subsystem, and said further processor-based system is programmed with instructions that cause the further processor-based system to implement said step of performing said verification procedure utilizing the encoded digital signature and the public key to determine whether the signature is valid.

15. The system as defined by claim 13, wherein said first and second finite fields are designated, respectively, as x-space and y-space, and further comprising said at least one processor-based subsystem being programmed with instructions to implement: generating a specified lattice in x-space and using said isomorphism to generate a corresponding lattice in y-space, and wherein said step of producing said digital signature includes initially producing a signature in the x-space lattice and then producing the digital signature in the corresponding y-space lattice.

16. The system as defined by claim 14, wherein said first and second finite fields are designated, respectively, as x-space and y-space, and further comprising said at least one processor-based subsystem being programmed with instructions to implement: generating a specified lattice in x-space and using said isomorphism to generate a corresponding lattice in y-space, and wherein said step of producing said digital signature includes initially producing a signature in the x-space lattice and then producing the digital signature in the corresponding y-space lattice.