您好,欢迎来到华佗小知识。
搜索
您的当前位置:首页基于DNA计算自组装模型的Diffie—Hellman算法破译

基于DNA计算自组装模型的Diffie—Hellman算法破译

来源:华佗小知识
第31卷第12期 计 算 机 学 报 VoI.31 No.12 Dec.2008 2008年12月 0F COMPUTERS CHINESE JOURNAL 基于DNA计算自组装模型的Diffie—Hellman算法破译 陈智华 (华中科技大学控制科学与工程系 武汉430074) 摘要DNA自组装计算模型是近年来引人关注的计算模型,已有基于自组装模型的二进制加法、乘法以及有限 域中的加法和乘法的讨论.文中利用DNA自组装模型设计的模乘系统,实现了素数P的本原根g连续乘方后模P 的数的排列,从而可以在线性时间内求解离散对数,为破译Diffie—Hellman密钥交换算法提供了新的生物方法.该 模乘系统使用了@(p)种自组装类型,组装的时间复杂度为O(p一1).系统最后组装结果提取出报告链后,经过 PCR和凝胶电泳读取离散对数结果.该模型扩展了DNA自组装计算模型的应用,为求取离散对数提供了新思路. 关键词DNA计算;DNA自组装模型;离散对数;整数排序;PCR TP301 中图法分类号Breaking the Diffie‘Hellman Key Exchange Algorithm in the Tile Assembly Model CHEN Zhi—HHa (Department of Control Science and Engineering,Huazhong University of Science&Technology.Wuhan 430074) Abstract Recent experiments have demonstrated that the simple binary arithmetic and logical operations can be computed by the self-assembly DNA tiles.This paper shows how the tile as— sembly process can be used to break Diffie—Hellman key exchange.In order to achieve this,the author used tile systems to construct the integers permutation from 1 to P一1 based on the truth table of the following numbers modular P:g rood P,g。rood P,…,gP~rood P.The output of the systems can be read by the standard sequence—reading operation that uses a combination of PCR and gel electrophoresis.Through the processes of tile systems,we can pick out the discrete log— arithm result which is the secret key in the Diffie—Hellman algorithm.So the key exchange by Diffie—Hellman algorithm is unsafe.The system can be carried out in@(P一1)assembly time with@( )tiles.The methods extend the applications of self-assembly mode1. Keywords tion;PCR DNA computing;DNA self—assembly model;discrete logarithm;integers permuta— l IntrOductiOn Adleman’s pioneering study[ 一。]set the stage sibility of molecular computing approaches to solve combinatorial problems.Bio—inspired computation could be used to solve complex mathematical prob一 1emsl_3L .With the vast parallelism and high densi— ty storage,DNA computing is employed to solve many comblnatoria1 optimization problems.Such an elegant idea inspired people to apply DNA corn— puting to cryptographic systems in such fields: breaking cryptosystems,encryption and decryp一 for the new field of bio—computing research.Using the vast parallelism to do the combinatoria1 search among a large number of possible solutions repre— sented by DNA strands,Adleman has solved some combinatorial search problems such as the Hamil— tonian path problem ̄1],and demonstrated the fea— 收稿日期:2008—04—18;最终修改稿收到日期:2008—10—15.本课题得到国家自然科学基金(60803113,60533010,60674106)和国家“八六三,, 高技术研究发展计划项目基金(2006AA012104)资助.陈智华,女,1976年生,博士研究生,讲师,研究方向为基于DNA计算的信息安全、 智能控制算法.E—mail;chenzhihua@mail.bust.edu.en. 12期 陈智华:基于DNA计算自组装模型的Diffie—Hellman算法破译 21l7 tion.hiding message and authentication。which re— quire an enormous amount of computation. the tile assembly model to implement arbitrary cir— cults[ .Similarly,Rothemund et a1.have imple— A number of proposals have been offered for mented a DNA tile system that could compute the X0R function and result in a Sierpinski trian— gle[ .breaking conventional eryptographic systems by DNA computing. For example,Boneh and Brun proposed theoretically the systems Adleman have proposed DNA computing meth— that could compute the sum and product of two numbers using the tile assembly model[1 7].He found that the adding and multiplying can be done ods to break the Data Encryption Standard(DES). Chang and Guo demonstrated that factoring the product of two large prime numbers could be com— pleted through biological operations using a molec— using@(1)tiles and the two computations can be carried out in linear time in tile assembly models. ular computer[ .Furthermore。this result indica— ted that the cryptosystems using public—key were perhaps insecure and also presented the clear evi— dence of the ability of DNA computing to perform complicated mathematical operations.Gehani pres— ented some procedures for DNA— based cryptogra—— phy based on OTP .Various modified DNA steg— anogram methods were proposed in order to im— prove the security.Leier presented two different cryptographic approaches based on DNA binary strands ̄ .The first approach hid information in DNA binary strands and the second approach de— signed a molecular checksum.Chen proposed a no— vel design of DNA——based molecular cryptogra—— phyL .They presented Carbon nano—tube based message transformation and DNA——based cryptosys—— tern.Clelland have demonstrated an approach of steganogram by hiding secret messages encoded as DNA strands among a multitude of random DNAs[”]. However,most of these proposals implemen— ted computing processes by performing a series of biochemical reactions on a set of DNA molecules. which require human intervention at each step. Thus,the difficulties of such methods for DNA computing are that the large numbers of laboratory procedures and the time consuming,which grow with the size of problems. 2 Tile Self-Assemble Model The tile self—assembly mode1.which is a for— mal model with the self-assembly molecules con— strained on a square lattice。is an extension for the tiling theory of Wang tiles[ that include a specific growth mechanism based on the physics of molecu一 1ar self—assembly.The first experimenta1 demon— stration of computation using DNA tile assembly was in[133.It gave a two layer,1inear assembly 0f TX tiles that performed a bit—wise cumulative XOR computation.Barish et a1.demonstrated a tile assembly system implementation that could copy an input hit and count it in binary[ .Cook et a1.used He then combined those systems to create two new systems with more complex behaviour to factor numbers and solve subset sum problems[ 一 . So DNA nanostruetures provide a nove1 meth— odology for us to resolve al1 kinds of mathematica1 and eombinatoria1 problems.Now the applications of DNA tiles are more and more universa1. 3 Breaking the Diffie—Hellman Algorithm in DNA Self-Assemble Models 3.1 Dirfie-Hellman Problem In the simplest form of the Diffie—Hellman scheme,everyone is assumed to know a large prime number P and one of its primitive root g.If two entities A and B wish to agree on a secret ran— dora value,then Step 1. A chooses a random element ∈ {1…P}and sends to B the value g rood P,and Step 2. B chooses a random element Y∈{1… P}and sends to A the value g mod P. The random value upon which A and B have agreed is g mod P,which both can calculate: A:A can calculate g mod P from z and g mod P via(g ) mod :g rood P,and B:B can calculate g orod P from Y and g rood P via(g y rood P—g rood P. The Diffie’—Hellman problem and its appliea—— tions have been well—studied in the world of corn— putational cryptography.It has even gained accept— ance in applied cryptography,and is used for key— agreement in such widespread protocols as SSH and TLS.Here,a molecular self—assemble model is proposed to search the secret elements.z and y. then the secret random value g rood P can be cal— ct11ated. 3.2 Models of Algorithmic Self-Assembly The abstract of tile self-assembly model, which provides a rigorous framework for analyzing algorithmic self—assembly,was originally proposed by Rothemund and Winfree[20_.The model consid— ers the assembly of rigid square objects or tiles. Elementary steps in the mode1 are simple assem— blies starting from a single seed tile and boosting 2118 计 算 机 学 报 2008拒 by the addition of single tiles,yet the formed structures may be surprisingly complex.Each unit of assembly is a square with glues of various types on each edge.The tile“floats”on a two dimen— sional plane and two tiles may stick when they col— lide if their abutting sides have compatible glues. Formally,a tile over a set of binding domains is a 4-tuple{ N,盯E, s, w)∈ indicating respec— tively the binding domains of the north,east, south and west sides.A position is an element of Z .The set of directions D一{N,E,S,W}is a set of four functions from positions to positions, i.e.Z to Z such that for all positions(.27,Y), N(x, )一(z, +1),E(x, )一(z+1, ),S(x, )一 (z,Y一1),W(z,Y)一( 一1, ).The positions (z, )and(1z ,y )are neighbors iff 3d E D such that d(x, )一(z ,Y ).For a tile t,for dED,we refer to bdd(£)as the binding domain of tile t on d’s side.According to this definition,tiles may not be rotated.A special tile empty一{null,null,null, “l1),represents the absence of all other tiles. The binding domains determine the interaction between tiles.A function g: to爬,where null E ,is a strength function denoting the strength of the binding domains,which may be 0,1,or 2 (respectively called null, weak, and strong bonds).If V , E ,then g(盯,盯 )一g( ,盯)and g(null,d)一0. Let T be a finite set of tile—types containing the empty tile.A configuration of T is a map from Z to T.For t E T,r is the configuration such that I1 (i,J)一t iff(i,J)一(z,y)and empty otherwise. A tile system is a quadruple S一(T,S,g,r), where T,g are as above and S is a set of super tiles called seed supertiles which assembly begins with, r is the temperature. If A is a configuration,then in the system S,a tile t can be attached to A at position(x,y)and produce a new configuration A iff: 1)(z, ) A, 2) dEDg(bdd(f),bdd一1(A( ( , ))))三三三r, 3)V(u, )EZ ;(u, )≠(z,y) A (u, )一 A(u, ),and 4)A (z, )一f. That is。a tile can be attached to a configura— tion only in empty positions and if the total strength of the appropriate binding domains on the tiles in neighboring positions meets or exceeds the temperature l-. Given a tile system S一(T,S,g,r),a set of tiles I1,and a seed configuration S:Z to r,if the above conditions are satisfied,one may attach tiles of T to S.Configurations produced by repeated at— tachments of tiles from T are said to be produced by S on S.If this process terminates,the configu— ration is called the fina1 configuration.If all possi— ble final configurations are identical,the tile sys— tem S is said to produce a unique final eonfigura— tion on S. 3.3 Modular Multiplication Using Self-Assemble DNA self~assemble methods could be used to search the random element z when we know the value g.P.and g .We construct a DNA self-as— semble system S=(T,S,g,r)to form the integers permutation from 1 to P’——1 based on the truth ta’— ble of the following numbers modular P:g mod P, g mod P,…,g _。rood P. A seed configuration S is designed,if condi— tions are satisfied,the tiles in the set T can be at— tached to S one by one unti1 the final configura— tions produce.The final configurations can be ex— tracted,and then according to the value g ,and gy.we extract the corresponding length DNA se— quence by PCR,the length j ust maps to the corre— sponding random element or Y.The system use @(1)types to construct I1 and 0(P)to construct T.The correctness of the system can be referred to the unique final configuration corollary , which states that if all the tiles in a system have unique east—south or west—south binding domain pairs,that system always produces a unique final configuration on some seed configurations. The set of r is designed as:I1一{S一<s,null, null,null>,E一<e,null,null,nul1),G一<g,null, null,nul1),SL一<null,1,s,nul1),ER一(null, null,e,1)).The 5 kinds of tiles are used to calcu— late the modular multiplication.Fig.1(a)shows a graphical representation of r,the value in the mid— die of each tile t represents that tile’s (t)value. These tiles with one binding domains,are denoted as bd,and let 6 N( )be the binding domain.The seed configuration for modular multiplication is shown in Fig.1(b),in which the primitive root g acts as continuous multiplier.Two special tiles are denoted as the symbols‘S’(with a strong bond, and three null bonds)and‘E’(with a weak bond, and three null bonds)respectively,are called boundary tiles,which are used to denote the start and the end of input numbers.They set limits on the extent of the calculation or patterning.and will facilitate a modular approach to the process.The strength of the binding domains for side with double line is set 2.These are the forms we store all inputs in our system. 12期 陈智华:基于DNA计算自组装模型的Diffie—Hellman算法破译 F 【 2119 S 图圈 g G (a) E g a S t 皿 ̄---Totle(…p-圆1)tiles h e C ( t Fig.2口is the faciend,b is the multiplier, :(口×6)roodp and the tile’s value口一“ n U o U S its top.The strength of the binding domains for side with double line is set 2.Fig.3(b)shows the m U t Fig.2 shows the calculation tile’s form used in p seed configuration S for the continuous multiplying {he primitive root g modular P.Fig.3(c)shows a sample of seed configuration which encodes the 一 3 mod 7.The start tile(boundary tile)which is described as the starting point is on the left side. Note that only one tile may be attached to this con— the system.In the calculation tile,the west side e r stores the value of faciend,the south side is the multiplier。and result of u一(a*6)rood P is stored in the east side.The strength of the binding do— mains for side with single line is set 1.The calcu— lation tiles arrange figures in calculating results, figuration because r一2.Fig.3(d)gives the final configuration for the example,and the report strand gives us the permutation:1×3 rood 7,1× 3 rood 7.….1×3 mod 7. which wil1 be explained in the following sections. Fig.3(a)shows the two boundary tiles used in the tile system and each tile’s name is written on b1 b2 母口 (a)boundary tiles (c)a sample of seed configuration for =3 mod 7 (b)seed configuration S (d)the final configuration Fig.3 Modular multiplication tile system Lemma 1.Let 一{1,2,…,P~1}and T be f5(1,一1)一E; the set of tiles defined by Fig.1(b),let g一1,r一2 and S be a seed configuration.Let Y—g rood P, here the prime number P determines the set’s scale lS(~P,一1)一S,S(一P,0)一SL; l Vi∈{1,…, 一1},s(一i,一1)一g; lFor all other(z, )EZ ,S(x,Y)=empty. It is clear that there is only one position where a tile may be attached to S,and after that tile atta— of T.Let S一(Jr,S,g,r),then,there must exist some(zo, 0)EZ such that:S(z。+1, 0—1)一 E,S(x0一P,yo一1):S,S(z0一P,yo)一SL;for ched there will also only be a single position where all iE{1,2,…,P一1},6 N(S(z。一i,y。一1)一g, and other positions(z, ),( , ) S.Then S pro— duees a unique final configuration F based on S and can gives us the permutation of modular P multipli~ a tile may be attached.By induction,for V£∈丁, the duple(bds(£),bdw(f)>is unique,and because r=2,it follows that S produces a unique final con— figuration on S.Let that configuration be F. cation,and the assembly time is@(P一1). V£∈T,the output is the product of its front tile’s output and the primitive root g modular Proof. Consider the tile system三={1,2,…, P一1}.Let r一{S一<s,null,null,null>,E一<e, null,null,null>,G一(g,null,null,null>,SL一 P,that is V1 i P一1,bdE(F(一i,0))一 (bd£(F(一i一1,0))×g)rood P.When the right boundary tile ER一(null,null,e,1>is attached to (null,1,s,null>,ER一<null, ull,e,1>}. Let the seed configuration S:Z +r: the position(1,O),the self—assemble process wil1 212O 计 算 机 学 报 2008伍 terminate.It indicates that S一(T,S,g,r)can pro— duce a unique final configuration on S and give the permutation of 1×g mod P,1×g rood P,…,1× g 一mod P.Because of r一2,only 3 tile with two sufficiently different from each—other.TAE tiles are used in our tile system.Fig.4(a)shows the abstract structure.Each tile is made up of six DNA strands.and there are six sticky ends that can complement and ligate together in the Watson— Crick sense.Each tile can have from one to six sticky ends.Non—used sticky ends can be bent into neighbors may attach at any time in this system. And then no tile may be attached to S until its left neighbor has.Thus the assembly time to give us the permutation is 一1 steps. 3.4 Extracting the Report Strand 口 loops so that they are not used as sticky ends. Note that,a single strand of DNA(shown in Fig.4(b))passes through each tile assembled and thus contains all integers stored as tile’s value. This single strand of DNA is called reporter To implement our method of modular multi— plication,we have to find suitable nucleotide se— quences for encoding all the symbols used in our computation.And we have to ensure that these se— quences(and their complementary sequences)are strand。and is also the result strand we need. TAE tile structure n (f1)TAEtile 三 The abstract of TAE tile \ (b)the structures of two TAE tile assembled Fig.4 Structures of two TAE tile assembled At the end of self-assemble process,ligase is To output the result of the computation we would use a modification of the standard sequence— reading operation that uses a combination of PCR and ge1 e1ectrophoresis.Using PCR we can obtain strands of different lengths representing the differ— ent exponent z in the result strand.According to the known values of the primitive root g and the added to seal the joints(see Fig.5)and then,the temperature of the chamber is increased to break up the tile structure into single strands.The result strand is then extracted by affinity purification using the complement of the‘RES’sequence.The ‘RES’nucleotide sequence would contain the site of a restriction enzyme at the end,so that the ‘RES’portion can be cut—off from the result strand by applying the appropriate restriction en— zyme.As a result we will get an ssDNA containing all the encoded integers in some permutation.This strand can then,be used in further computations. The report strand passing through all tiles result value Y,we start PCR.The part of report strand“g—y’’can be amplified by PCR method by putting the“g”as a forward primer and“Y”as a reverse primer.The needed string could be gel pu— rified since its length would be different from the remaining sequence.The corresp0nding needed string“1×g mod P,1×g rood P,…,y一1 X g rood P”can be extracted,see Fig.6(a). As we know,the security of the Diffie—Hell— man key exchange protocol lies in the fact that, While it is relatively easy to calculate exponentials (a)the final configuration:F contain the site of a restrict’ion enzyme modulo a prime。it is very difficult to calculate dis— crete logarithms.Here is an example,key ex— change is based on the use of the prime number (b)the extracted report strand 户一7 and a primitive root of P,in this case g一3. Fig.5 Extracting the report strand The entities A and B select secret keys XA and XB, respectively.Each computes his or her public key: 12期 陈智华:基于DNA计算自组装模型的Diffie—Hellman算法破译 2121 (a)the needed strands by PCR 2000,6 1000,5 750,4 500,3 250,2 1O0.1 (db,X) (b)read out X by gel electrophoresis Fig.6 Read Out X by gel electrophoresis after PCR YA一5 and YB一6.After they exchange public keys,each one can compute the common secret key:K一(yB) mod P一(YA) orod P.In fact, we could get the P,g, and YB easily.We use the DNA tile systems and biological technologies to get the XA and XB.In this case,P一7,g===3 and YA一5,the part of report strand“3—5”can be am— plified by PCR method by putting the“3”as a for— ward primer and“5”as a reverse primer.The nee— ded string could be gel purified since its length would be different from the remaining sequence。 see Fig.6(b).Based on the relation of sequence 1ength and z,we can read out the XA一5.Then P一7,g一3 and YB一6,we repeat the same pro— grams and read out the XB一3.So we can compute their common secret K一6 rood 7—5。mod 7—6. Through the tile system,PCR and gel electro— phoresis,we can break the Diffie—Hellman algo— rithin。 4 Conclusion DNA self assembly is expected to be usefu1 in information security.Our method for breaking Dif- fie—Hellman algorithm extends the technique used by LaBean et a1.Ez2]for binary addition and XOR. The advantage of our methods is that once the ini— tial strands are constructed.each permutation op— eration is processed very fast through the self— as—+ sembly and the output can be directly passed as in— put to another computation.The only time consu— ming operation is the reading of the output. The permutation of 1×g mod P.1×g2 rood P.…,1×g 一rood P can be achieved by the self— assembly system.The tile systems use@( )input tiles and compute in@(P)steps.We extract the report strand in the final configuration,then use PCR and gel electrophoresis to read out the dis— crete logarithms.Furthermore,we can compute the common secret key used in Diffie—Hellman scheme. The bottleneck of the self—assembly system is that the types of basic input tiles is linear correla— tion with the prime number P.The needed input tile types will increase with increasing P.When P is large enough,it is very difficult to find enough distinct types tile to carry out computation.If the improved tile system can break the bottleneck,the system would be more meaningfu1. The nanotechnology and biological technology hold tremendous promise.But many technical hur— dles will have to be overcome before complex algo— rithmic self—assembly can be developed into a prac- tical commercial technology.If the molecular word can be controlled at wil1.it may be possible to achieve vastly better performance for information storage and information security. References Adleman L M.Molecular computation of solutions to combi— natorial problems.Science,1994,266:1021 1024 [2] Adleman L M.Computing with DNA.Scientific American. 1998,279(2):54-61 [33 Liu L Q.1 iu G W,Xu J.Liu Y C.‘Solid phase based DNA solution of the coloring problem.Progress in Natural Sci— ence,2004,14(5):104—107 Pan L Q.Carlos M V.Solving multidimensional 0-1 knap— sack problem by P systems with input and active membranes. Journal of Parallel and Distributed Computing,2005,65 (12):1578 1584 Boneh D,Dunworth C,Lipton R J.Breaking DES using a molecular computer//Proceedings of the l st DIMACS Work— shop on DNA Based Computers.1995 1 37—65 E6] Adleman L M,Rothemund P W K,Roweis S,Winfree E. On applying molecular computation tO the data encryption standard//Proceedings of the 2nd Annual Meeting on DNA Computers.1996:1O一12 [7] Chang W L.Guo M Y,Michael S H.Fast parallel molecular algorithms for DNA—based computation:Factoring integers. IEEE Transactions on Nanobioscience,2005,4(2):149-163 Gehani A,LaBean T H,Reif J H.DNA—based cryptogra— phy//Proceedings of the 5th DIMACS Workshop on DNA— Based Computers,MIT.1999:233 [9] Leier A,Richter C,Banzhaf W,Rauhe H.Cryptography with DNA binary strands.Biosystems,2000,57:13—22 2122 计 算 机 学 报 2008钲 [1O] Chen J.A DNA—based biomolecular cryptography design// Proceedings of the 2003 International Symposium on Circuits and Systems:ISCAS 2003,2003:822 [11] Clelland C T,Risca V,Bancroft C.Hiding messages in DNA microdots.Nature,1999,399:533—534 [123 Wang H.Proving theorems by pattern recognition I.Bell System Technica1 Journal。1961,4O:卜42 [13] Mao C,LaBean T H,Reif j H,Seeman N C.Logical com— putation using algorithmic self-assembly of DNA triple——cross—_ over molecules.Nature,2000,407:493-496 [14] Barish R,Rothemund P,Winfree E.TWO computationa1 primitives for algorithmic self—assembly:Copying and count— ing.Nano Letters,2005,5(12):2586—2592 [15] Cook M,Rothemund P,Winfree E.Self-assembled circuit patterns//Proceedings of the 10th International Meeting on DNA Computer.Milan,Italy,2004:91-107 [i63 Rothemund P,Papadakis N,Winfree E.Algorithmic self—as— sembly of DNA sierpinski triangles.PLoS Biology,2004,2 (12):2041—2053 CHEN Zhi—Hua, born in 1976, Ph.D.candidate。lecturer.Her research interests include DNA—based information security and intelligent control algo— rithm. [17] Brun Y.Arithmetic computation in the tile assembly model: Addition and multiplication.Theoretical Computer Science, 2006,378(1):17~31 [18] Brun Y.Nondeterministic polynomial time factoring in the tile assembly mode1.Theoretical Computer Science,2008, 395(1):3—23 [19] Brun Y.Solving NP—complete problems in the tile assembly mode1.Theoretical Computer Science,2008,395(1):3卜46 [2o3 Rothemund P,Winfree E.The program—size complexity of self-assembled squares//Procee dings of the 32nd Annual ACM Symposium on Theory of Computing(STOC).Port— land,200l:459—468 [21] Rozenberg G,Spaink H.DNA computing by blocking.The— oretical Computer Science,2003,292(3):653-665 [223 LaBean T H.Winfree E,Reif J H.Experimental progress in computation by self—assembly of DNA t|ljngs//Pr0ceedings of the 5th International Meeting Oil DNA Based Computers (DNA5).MIT,Cambridge MA,1999:123—140 

因篇幅问题不能全部显示,请点此查看更多更全内容

Copyright © 2019- huatuo0.cn 版权所有 湘ICP备2023017654号-2

违法及侵权请联系:TEL:199 18 7713 E-MAIL:2724546146@qq.com

本站由北京市万商天勤律师事务所王兴未律师提供法律服务