CN1391413A - Method for implementing genetic programming of encoder to generate sequence with pointed performance - Google Patents
Method for implementing genetic programming of encoder to generate sequence with pointed performance Download PDFInfo
- Publication number
- CN1391413A CN1391413A CN 02138011 CN02138011A CN1391413A CN 1391413 A CN1391413 A CN 1391413A CN 02138011 CN02138011 CN 02138011 CN 02138011 A CN02138011 A CN 02138011A CN 1391413 A CN1391413 A CN 1391413A
- Authority
- CN
- China
- Prior art keywords
- encoder
- fitness
- value
- criterion
- colony
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
A genetic programming method for generaitng the encoder with particular performance features that a series of operators and operands are used to obtain N equations which are just the N encoders, where N is group scale numeral, the adaptivities (or evaluation index) of N individuals in a group are calculated out, where the encoder is dependent to the its evaluation index defined by user, the genetic probability of each equation to next descendant is calculated out, the genetic variation operation of the group is performed to derive the next-descendant group, and after deriving a certain number of descendants, the equation with the highest adaptivity is used as the final result.
Description
Affiliated field
The present invention relates to a kind of genetic programming implementation method of encoder to generate sequence with pointed performance.
Background technology
Various random sequences are at mathematics, and engineering and other scientific domain widely have very high using value.As generation, spread spectrum communication and the 3G (Third Generation) Moblie of Monte Carlo stochastic simulation, decision theory, information encryption, lottery number, satellite communication, optical communication technique etc.
And in the mobile communication technology of a new generation,, having adopted CDMA (Code Division Multiple Access) in order more effectively to utilize frequency resource, this just needs performance to satisfy the multiple access orthogonal code of particular requirement well.Equally in other respects, such as optical communication field, require the performance of sign indicating number and other communication mode different again, we just need to generate the method for this type of yard again.
So, find a kind of comparatively method in common, it is significant solving this class problem of sequence that generates certain performance preferably.The present invention has found a kind of logical method that can solve this type of problem preferably, has disclosed a kind of approach that generates this type of encoder effectively.
In the present existing technology, the method that generates certain particular sequence generally is to obtain by utilization number theory theory, or uses algebraical theory to push away.Such as pseudo random sequence, it usually obtains by deterministic algorithm, and this algorithm to be the programmer know in advance, this conventional method needs the seed of random number simultaneously, so the random number of output is reproducible.Therefore performance is bad at random for the number that obtains by deterministic method.For another example, have the human nonlinear kinetics to generate spreading code, its performance generally can not reach capacity.
Generally speaking, the major defect of these methods is:
(1) performance generally can not be approached its limiting value;
(2) number of codes of Sheng Chenging is limited;
(3) to solve a problem, can only tailor the method for a special use, not have versatility.
Summary of the invention
The present invention is for solving existing encoder generated code limited amount, performance generally can not be approached its limiting value, self only being confined to does not have the problem of versatility in the method suggested scope, and a kind of genetic programming implementation method of encoder to generate sequence with pointed performance is provided.Genetic programming (Genetic Programming) is based upon on the basis of natural selection rule of Darwin's survival of the fittest, utilizes variation and duplicate to generate algorithm, and this algorithm can spontaneously be created the computer program of updating.
Technical scheme of the present invention is:
(1) form N formula at random with sequence of operations symbol and operand, each formula is exactly an encoder in fact, and wherein N is called the population size number;
(2) each encoder is had an evaluation index, just fitness value is calculated N individual fitness value in the colony, and its evaluation index is specified by the user, and has what evaluation index will generate what encoder;
(3) based on fitness value, the criterion that is directly proportional with fitness value according to genetic probability is calculated each formula and is genetic to follow-on probability.Then this colony being carried out hereditary variation operation develops and colony of future generation.Certain algebraically that develops as required, with fitness is the highest in the final colony formula as final result.
Two embodiments of the present invention are: 1) its generates the complex sequence encoder under Shannon meaning, and the fitness criterion that its adopts is the Shannon criterion; 2) the multiple access orthogonal code among the generation CDMA, the fitness criterion that it adopts is: find out the maximum y in auto-correlation and the cross correlation value in a code character, if y＞a, fitness value=0 (a is an adjustable value), else fitness value=Shannon entropy-y (number of control 0).
Owing to adopted genetic programming method, made method of the present invention search volume scope very big, thereby guaranteed that the result has good performance with hereditary variation function.
Because the intrinsic stochasticity of genetic programming makes by each operation, can obtain the outstanding and mutual dissimilar encoder of performance.
Because the fitness index can be specified by the user, make of the present invention being of wide application.
Because adopting Shannon entropy is the fitness criterion, can obtain the most complicated encoder under the information-theoretical criterion, so this encoder has very big application prospect at aspects such as information encryption, Monte Carlo stochastic simulation, lottery number generations.
Owing to adopted in the mobile communication requirement to the multiple access orthogonal code as the fitness criterion, can obtain good multiple access orthogonal code sets.
Because the high degree of adaptability and the versatility of the method make the present invention in the wide spectrum of communication very big using value be arranged, as the generation of various sign indicating number types in satellite communication and the optical communication etc.
Description of drawings
The invention will be further described below in conjunction with the drawings and specific embodiments
Fig. 1 is the programme tree structure of pseudo-random sequence generator of implementation method example of the present invention
Fig. 2 is the programme interlace operation schematic diagram of implementation method of the present invention.(a) for before the interlace operation.(b) for after the interlace operation.
Fig. 3 is the present invention's encoder output sequence Shannon entropy that implementation method produces change curve along with genetic algebra of programming
Fig. 4 is the programme autocorrelation performance curve of 128 random sequence encoder output sequences that implementation method produces of the present invention under the Shannon criterion
Fig. 5 is the programme multiple access orthogonal code autocorrelogram of implementation method of the present invention
Embodiment
The present invention is from the genetic programming technology, encoder is regarded as one section program, and the side that adopts hereditary variation is through testing the encoder of using the genetic programming method to develop and generate, its output sequence not only performance approaches the best, and every operation one secondary program can obtain a new encoder.Should be noted that especially, the present invention is to provide a kind of logical method, its flexibility is that the performance index of encoder are by user's appointment.Be the arthmetic statement of this method below:
The first step is determined to use what symbol sebolic addressing, what rule to constitute an encoder by.Because our desirable encoder output can only be 0 and 1, encoder output can only be the integer number.Therefore, associate following rule:
(1) encoder realizes that by a binary tree node is divided into two classes: operand node, operator node (binary operator).
(2) operator be only limited to+,-, *, DIV, MOD}, wherein DIV is divided exactly operator.And, owing to consider special circumstances, thus regulation n MOD 0=1, n DIV 0=1, so no matter when, evolutionary process all can go on.
(3) operand be only limited to 0,1,2,3, j}, wherein j is the input of encoder, but value is the nature manifold, when j just obtains a sequence when 1 gets N, though this sequence had more than by 0 when beginning to develop, 1 forms, but develops to certain algebraically, just can only be exported 0,1 encoder.Result (15 generation) such as me is (the preorder traversal sequence of binary tree): M++Dj+MDj2++MDj+13+Mj2MD+312-*0103M-+j31+++D0j*113M3-Dj2 13332
Be drawn as tree type such as Fig. 1 (because paper is not enough, the part of can only drawing).
Because we will carry out hereditary variation operation to an encoder colony, so at first we must obtain M encoder (M is the population of colony) at random, this be need solution a problem.
In second step, determine the fitness of body one by one.This judgment criteria is to be provided according to their needs by the user.We are that example illustrates now with the Shannon entropy.
To each tree type encoder, if do not import node J, this encoder output is constant so, and we it will be appreciated that this encoder output sequence Shannon entropy is zero.If input node J is arranged, allow so J from 1 continuous value to M (being taken as 5000 my program), just obtain a series of output valves, these values are stored in order.
I have used Shannon criterion (also available certainly other criterions): to deciding the sequence of word length (h), have 2
hPlant possibility, when the probability of each sequence appearance equates, the Shannon entropy maximum, size is h, i.e. formula E
h=-∑ p
HiLog
2p
HjReach maximum h.Because we will optimize all sequences of word length less than k (k gets 7 in this program), so the definition fitness function is
Be not difficult to know E
TotalThe shannon limit value be
。So a given code tree is just obtained its fitness.Select by the individuality of heredity according to the rule of " individual probability by heredity is directly proportional with its fitness " then.Along with the increase of algebraically, the Shannon entropy of the output sequence of the encoder that obtains can approach limiting value, that is to say, encoder can approach optimum under the Shannon meaning.(during in fact to the 3rd generation, the output sequence entropy of the encoder that obtains has just reached more than 25, and after this along with the increase of algebraically, entropy approaches limiting value 28 with very slow speed)
In the 3rd step, intersect and mutation operation
Obtain in the colony behind each individual fitness, will carry out a hereditary variation and operate and obtain colony of future generation.At first and since the fitness sign body is to the adaptedness of environment one by one, the big individuality of fitness adapts to environment, it should easier existence and is genetic among the next generation and goes.And can bring a problem like this, the individuality that fitness is big might be by heredity, and the little individuality of fitness also might be by heredity (realizing by random number when realizing because programme), in order to keep the stability of algorithm, I make fitness come preceding 10% individual unalterable copying among the next generation, and the hereditary variation operation is only carried out 10% later individuality.
During the hereditary variation operation, elder generation is selection criterion as described above, selects a pair of parents (parents), that is to say and has selected two trees, and these two individualities will carry out mating so, and process very clearly is presented among Fig. 2
The position that it should be noted that the crosspoint that Fig. 2 (a) arrow shows is also selected at random, and the probability selection operator node with 0.9 is counted node with 0.1 probability selection operation.In addition, owing to will reduce the running time of program, to guarantee during interlace operation that two tree degree of depth that obtain are no more than 15.In order to enlarge the range of choice of algorithm, need to carry out mutation operation sometimes, as the occurring in nature of reality, the probability that biological sudden change takes place is very little, so stipulated in my program that the individual probability that morphs is 0.1.
Following mask body is discussed the specific implementation of a lower variation: because node has two types in the one tree, and operator and operand, variation not hard to imagine so has two kinds of possibilities:
A. operator → operand
B. operand → operator
To situation A, originally left subtree under the operator node and right subtree will be deleted fall.The operator node that variation obtains must newly grow left and right sides subtree, otherwise is exactly nonsensical tree to situation B! And, stipulate only to grow in this situation two operand nodes owing to will reduce the scale of program running.Further, the left and right sides subtree that grows among the selection of variation node and the situation B also has randomness.
Use the operation result of this technical scheme
As above said, among the present invention, the fitness index of encoder is by user's appointment, and the user can oneself change as required.Below under the Shannon entropy criterion and at 3G (Third Generation) Moblie two specific embodiments under the requirement of multiple access code are being provided operation result respectively.
Specific embodiments one: with the Shannon entropy is the operation result of fitness criteria.
Figure 3 shows that in running that (1) the output sequence entropy of the best individuality of performance is along with the change curve of genetic algebra in the colony.As seen, convergence of algorithm speed is very fast.
(2) below be and same once 128 random sequence generators that obtain in service of Fig. 3 (the blue formula of middle preface head sea):
D1M-M+M+DM-2j2*Mj31Mj3j*13D-j13+DjM+M+*+jj1+M+-2jMj3j+j1+jM1+M*Mj311*2j+M+*33Mj3jDjM*13M-2jD+Dj3Dj33*Dj3M32j+M212
The output sequence Shannon entropy of this encoder reaches more than 27.9, this encoder approaches the most complicated under the Shannon meaning, and have only about 22 with the randomizer Shannon entropy that Matlab software provides, and the Shannon entropy of the m sequence that produces with shift register is less than 20.
(3) Figure 4 shows that the output sequence autocorrelation performance curve that utilizes 128 random sequence encoders of genetic programming realization once in service together with Fig. 3.
Specific embodiments two: with the requirement to coding in the 3G (Third Generation) Moblie is the operation result of fitness criteria.
Flexibility of the present invention is that to a great extent it can generate the encoder that satisfies various criterion according to user's different needs.Need the very little encoder of output sequence auto-correlation secondary lobe such as us, this encoder is widely used in the communication technology, mainly due to the needs of anti-multipath time delay in the mobile communication.
The target of the communication technology is to realize personal communication, but radio frequency resources is limited, transmission information as much as possible in limited frequency band how, and the solution of this technical problem is the performance of encoding to a great extent.In CDMA technology, by different signals with mutually orthogonal code character spread spectrum, then these signals are sent in same frequency range.These signals are aliasings when not decoding, even are submerged in the noise.Receiver can be selected the signal that uses the special code type with correlator in the signal of aliasing then.Many users of many users of same sub-district and neighbor cell are operated on the same frequency in cdma system, thereby it is the most effective a kind of multi-access mode on the service efficiency of frequency, but the interference of the multiple access among the CDMA still plays restrictive function to the capacity of system, because along with being on the increase of while active user, the multiple access interference level is inevitable increasing, when being increased to a certain degree, the signal to noise ratio of received signal will become and can't stand.
So design the orthogonal code sets that generates certain-length, become very important.Yet consider to exist serious multipath effect in the radio communication channel, so we need the good orthogonal code sets of autocorrelation performance.This like this sign indicating number both can resist multiple access to disturb, again can anti-multipath time delay, the greatly performance of improvement communication.
In fact just because of the incomplete quadrature between the group code, the far and near interference arranged, if the cross-correlation of sign indicating number is zero, blocking the base station except too close base station between the different address codes receives, other users are not caused interference, just do not need good power control, system will simplify many.
If the cross correlation of intersymbol the time is zero in alignment only, differ one just non-vanishing, this has just limited the performance of SCDMA, because there is multipath transmisstion in the wireless channel.
To sum up, we need a kind of like this multiple access code character of performance:
(1) the auto-correlation secondary lobe of this class sign indicating number is pressed as far as possible for a short time
(2) wantonly two group codes in the code character, cross-correlation is very little.
(3) Ma density is high as far as possible.
The fitness index can be defined as like this: find out in a code character in auto-correlation and the cross correlation value
Big value y, if y＞a fitness value=0 (a is an adjustable value)
Else fitness value=Shannon entropy-y (number of control 0)
With 64 code characters is example, provides an orthogonal code sets.00000000001000000000000001100000000000000000100100000000100000 (1)0000000001000000000000001100000000000000000001001000000001000000 (2)0000000010000000000000011000000000000000000010010000000010000000 (3)
Figure 5 shows that the autocorrelogram of certain multiple access orthogonal code.This code character ISN is controlled between 0,1 with the cross correlation value of sign indicating number.
Certainly our code length that needs is much longer in actual applications, reaches kilobit and even more than the myriabit.In the optical communication even need the low-density code of ten myriabits.As long as computer is enough fast, such as parallel machine, the present invention also can be competent at.
Claims (3)
1, the method with genetic programming generates the encoder that satisfies particular requirement, it is characterized in that in the following ways:
(1) form N formula at random with sequence of operations symbol and operand, each formula is exactly an encoder in fact, and wherein N is called the population size number;
(2) encoders have an evaluation index, and just fitness value is calculated N individual fitness value in the colony, and its evaluation index is specified by the user, and has what evaluation index will generate what encoder;
(3) based on fitness value, the criterion that is directly proportional with fitness value according to genetic probability is calculated each formula and is genetic to follow-on probability.Then this colony being carried out hereditary variation operation develops and colony of future generation.Certain algebraically that develops as required, with fitness is the highest in the final colony formula as final result.
2, encoder generation method according to claim 1 is characterized in that generating the complex sequence encoder under the Shannon meaning, and the fitness criterion that its adopts is the Shannon criterion.
3, claim 1 described encoder generation method, it is characterized in that generating the multiple access orthogonal code among the CDMA, the fitness criterion that it adopts is: find out the maximum y in auto-correlation and the cross correlation value in a code character, if y＞a, fitness value=0 (a is an adjustable value), else fitness value=Shannon entropy-y (number of control 0).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 02138011 CN1391413A (en) | 2002-07-26 | 2002-07-26 | Method for implementing genetic programming of encoder to generate sequence with pointed performance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 02138011 CN1391413A (en) | 2002-07-26 | 2002-07-26 | Method for implementing genetic programming of encoder to generate sequence with pointed performance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1391413A true CN1391413A (en) | 2003-01-15 |
Family
ID=4749234
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 02138011 Pending CN1391413A (en) | 2002-07-26 | 2002-07-26 | Method for implementing genetic programming of encoder to generate sequence with pointed performance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1391413A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101335538B (en) * | 2008-06-18 | 2012-06-27 | 清华大学 | Method and apparatus for spread-spectrum sequence generation |
| CN102187656B (en) * | 2008-10-27 | 2014-10-01 | 日本电信电话株式会社 | Pixel prediction value generation procedure automatic generation method, image encoding method, image decoding method, devices using these methods, programs for these methods, and recording medium on which these programs are recorded |
| CN104595924A (en) * | 2015-01-14 | 2015-05-06 | 河北省电力建设调整试验所 | Method and device for establishing boiler combustion process model |
-
2002
- 2002-07-26 CN CN 02138011 patent/CN1391413A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101335538B (en) * | 2008-06-18 | 2012-06-27 | 清华大学 | Method and apparatus for spread-spectrum sequence generation |
| CN102187656B (en) * | 2008-10-27 | 2014-10-01 | 日本电信电话株式会社 | Pixel prediction value generation procedure automatic generation method, image encoding method, image decoding method, devices using these methods, programs for these methods, and recording medium on which these programs are recorded |
| US9124289B2 (en) | 2008-10-27 | 2015-09-01 | Nippon Telegraph And Telephone Corporation | Predicted pixel value generation procedure automatic producing method, image encoding method, image decoding method, apparatus therefor, programs therefor, and storage media which store the programs |
| CN104595924A (en) * | 2015-01-14 | 2015-05-06 | 河北省电力建设调整试验所 | Method and device for establishing boiler combustion process model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1218526C (en) | Scrambler/descrambler and method theirof | |
| CN1496038A (en) | Method for producing scramble code in mobile communication system | |
| CN1384635A (en) | Pseudo-noise state generating device and method | |
| CN1365559A (en) | Soft base digital receiving adaptable to various multiplexing systems | |
| CN109039522B (en) | Chaos sequence-based method for optimizing balance of spreading codes | |
| Rowe et al. | Properties of gray and binary representations | |
| Gorbenko et al. | Cryptographic signals: requirements, methods of synthesis, properties, application in telecommunication systems | |
| CN1426635A (en) | Post processing of spreading codes in mobile telecommunications system | |
| WO2003049226A1 (en) | Method for assigning variable-length walsh codes for code division multiple access communications systems | |
| CN1391413A (en) | Method for implementing genetic programming of encoder to generate sequence with pointed performance | |
| CN1753398A (en) | Enhanced method for demodulating constant-amplitude multi-code biorthogonal modulation signals | |
| Wang et al. | A Chaotic Key Expansion Algorithm Based on Genetic Algorithm. | |
| CN108696327A (en) | Large size mixing P-H is orthogonal, constructive method of class orthogonal matrix | |
| Wu et al. | A parallel interleaver design for IDMA systems | |
| Chandra et al. | Small set orthogonal Kasami codes for CDMA system | |
| Li et al. | Extended Reed-Solomon codes for optical CDMA | |
| CN1356799A (en) | Method for generating orthogonal spectrum-spreading code in mobile communication system | |
| CN1137549C (en) | Design method for 2-D signals with low/zero correlation region | |
| CN1251435C (en) | Method of generating bidimensional orthogonal variable spread spectrum coefficient code in multi-carrier wave direct sequency CDMA cemmunication system | |
| CN1166073C (en) | Random extending code selection method reaching shannon limit | |
| CN1538629A (en) | Apparatus for mapping and spreading data symbols in mobile communication systems | |
| CN102324996B (en) | Method and device for generating multi-value pseudorandom sequence | |
| CN1287425A (en) | Multiphase orthogonal spectrum spreading code design and its spread-eliminating method | |
| CN1694373A (en) | Method for solving multidress interference in TD-SCDMA system | |
| CN1171408C (en) | Method and device for generating real-time channel code sequence in WCDMA system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |