CN101616074B - Multicast routing optimization method based on quantum evolution - Google Patents
Multicast routing optimization method based on quantum evolution Download PDFInfo
- Publication number
- CN101616074B CN101616074B CN2009100222896A CN200910022289A CN101616074B CN 101616074 B CN101616074 B CN 101616074B CN 2009100222896 A CN2009100222896 A CN 2009100222896A CN 200910022289 A CN200910022289 A CN 200910022289A CN 101616074 B CN101616074 B CN 101616074B
- Authority
- CN
- China
- Prior art keywords
- node
- quantum
- destination node
- multicast tree
- multicast
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a multicast routing optimization method based on quantum evolution, relating to the evolutionary computation field and aiming to solve the problems of long optimization time, easy convergence to local optimum, and the like existing in the prior art. The multicast routing optimization method based on quantum evolution comprises the following steps: 1. generating a stochastic network and setting operational parameters; 2. solving all alternative paths satisfying time-delay conditions for each destination node; 3. carrying out quantum coding on the alternative path set of each destination node so as to acquire a state matrix; 4. carrying out quantum observation on the state matrix so as to obtain a set of binary strings; 5. decoding the binary strings so as to obtain the selected path of each destination node, and computing a multicast tree fitness function; 6. carrying out quantum variation on the state matrix by a quantum revolution door; 7. obtaining a set of new binary strings by observing the state matrix, and then computing a new multicast tree fitness function after the new binary strings are decoded. The invention has low complexity of computation and low cost of the optimized multicast tree and can be used for effectively allocating network resources.
Description
Technical field
The invention belongs to network service and calculate the field, relate to a kind of multicast route method, can be used for improving the service quality of internet.
Background technology
From at the end of last century to the beginning of this century be the epoch of network technology great development, particularly be that the IP network of representative is maked rapid progress especially with the internet.Number of users on the internet continues to be explosive increase, and online application is turned to application such as FTP, WWW by traditional Email.In addition, based on the new application of internet and business also constantly releasing, as ecommerce, IP phone, video conference etc.But, satisfy these professional demands, particularly to guarantee the specific demand such as bandwidth, time delay of some real time business, be difficult to finish with the service of doing one's best in the present internet.The characteristics of the service of doing one's best are to provide data transport service of the same race to all application, resource shortage to network effectively distributes and management, when offered load is light, the transmission service quality that each application obtains still can, but along with the increasing of number of users, the load of network also will increase, at this moment, the behavior of various application shows as competition network resource disorderly, and that causes Internet resources unreasonablely takies various application and respectively be its profit, and consequently service quality worsens mutually.Present technique of internet is badly in need of improving so that efficient resource allocation and management to be provided.Foregoing network application, characteristics as network applications such as video conference, interactive entertainment, sound/visual telephone, real-time multimedia broadcast, Distributed Calculation, video request program and remote teaching relate to the mutual of a plurality of members, have the feature of multicast in essence.
KPP is a limited multicast route method of time delay that early proposes, it has used for reference the thought of KMB method, at first construct a complete graph that only comprises the set of source node and destination node, satisfy the expense shortest path of delay constraint among the corresponding former figure in every limit in this complete graph between two nodes: from complete graph, produce a limited generation tree of time delay then; The limit that will generate tree at last replaces with the limited minimum cost path of the time delay among the former figure, and removes the loop of generation.The shortcoming of this method is: the complexity height; When separating when existing, it may can not find optimal solution.
The BSMA method is a kind of improvement to the KPP method, and it adopts the feasible solution that finds problem earlier, then feasible solution is improved, and makes its performance more near the method for optimal solution, and detailed process is: generate multicast tree with the LD method earlier; Littler and super limit that satisfy delay requirement replaces generating the bigger super limit of expense in the tree with expense then, when the high super limit of replacement expense, used the k shortest-path method, wherein super limit be meant initial sum end node all degree of being more than or equal to a paths of 2 node, super limit also can be the paths between two destination nodes, or degree is greater than 2 node and the paths between the destination node; Carry out this step repeatedly, till the expense of whole generation tree can not reduce again.The shortcoming of this method is the solution procedure complexity, is not suitable for the optimization large scale network.
Propose recent years based on the solution multicast path of genetic method by method, though strengthened ability of searching optimum than classical way, it is absorbed in " precocity " easily, is difficult to obtain optimum multicast tree.And based on artificial immunity theoretical solve multicast path by method, need can keep the diversity of population by continuous Cologne population, obtain optimum multicast tree, but make the ageing of method reduce greatly like this.
Summary of the invention
The objective of the invention is to overcome the deficiency of prior art, propose a kind of multicast route method, find the solution the multicast tree time reducing, obtain less multicast tree cost, reduce the assorted degree of operation method, improve the ageing of network based on quantum evolution.
Realize that technical thought of the present invention is: regard multicast routing problem as combinatorial optimization problem, make the minimized combined sequence of multicast tree cost as optimal solution with the quantum calculation search; Utilize quantum calculation global convergence fast, search constitutes the optimal solution of multicast tree problem, approaches the performance of optimum multicast tree service.The specific implementation step is as follows:
(1) generates random network, and source node, destination node are set, given operational factor and maximum allowable delay in network;
(2) each given destination node is found the solution the alternative path collection that all satisfy the time delay condition, and arrange by ascending order;
(3) to preceding 2 of each destination node
NBar (N ∈ [1,4]) alternative path collection carries out quantum coding, obtains a state matrix;
(4) to representing the state matrix q of each destination node alternative path collection coding
tCarry out quantum observation, obtain one group of binary string;
(5) above-mentioned binary string is decoded, obtain the selected path of each destination node, through constituting a multicast tree, and calculate the fitness function of this multicast tree by these routings; Fitness=1/MultiTreeCost, wherein MultiTreeCost is the multicast tree cost;
(6) to representing the state matrix q of each destination node alternative path collection coding
t, carry out the quantum variation by the quantum revolving door;
(7) state matrix to each destination node alternative path collection coding after the expression variation carries out quantum observation, obtains one group of new binary string;
(8) new binary string is decoded, obtain the new routing of each destination node footpath, by these new routings through constituting a multicast tree, and calculate the fitness function of this tree, if new multicast tree fitness function is better than the fitness function in the step (5), then adopt the optimum multicast tree building mode that searches by the mode conduct of these new routings through constituting multicast tree, otherwise the optimum multicast tree building mode that continues to adopt the building mode conduct in the step (5) to search;
(9) judge whether to satisfy end condition,, finish to optimize and adopt the optimum multicast tree building mode that finally searches, information is sent to different destination nodes with parallel mode along this branch if satisfy end condition, otherwise, continued to carry out (5) step.
The present invention compared with prior art has following advantage:
1, computation complexity is low
Suppose that network size is n, the destination node number is m, and the alternative path collection is P, and population scale is M, and the method iterations is g, and the complexity of then seeking alternative path is O (n); The complexity of alternative path being carried out the ascending order arrangement is O (log (P)); The time complexity of seeking optimum multicast tree be O (g * M * m), so time complexity of the present invention is approximately:
O(g×M×m)+O(log(P))+O(n)
By abbreviation, the total time complexity of the present invention is approximately:
O(n
2×m)
But the time complexity of existing BMSA method is O (n
3* logn), the present invention is better than BMSA from time complexity.
2, the emulation experiment performance is good, and it is little to try to achieve the multicast tree cost under the equal conditions
In order to verify the superiority of quantum evolution multicast route method, the present invention compares itself and traditional genetic method, Immune Clone Selection method.Emulation experiment shows with the Immune Clone Selection method to be compared, the quantum evolution multicast route method has obviously reduced time complexity, compare with traditional genetic method, having better search capability. this is because of the concurrency that has effectively utilized in the quantum calculation principle, and in each iterative process, the quantum evolution multicast route method is not certain paths of choosing at random, but obtain by the observation state matrix, make full use of the information of parent, accelerate convergence of algorithm, obtained optimum multicast tree cost.
Description of drawings
Fig. 1 is an optimization FB(flow block) of the present invention;
Fig. 2 is that the present invention seeks the alternative path flow chart;
Fig. 3 is Δ θ
iSelection strategy figure;
Fig. 4 is the performance comparison diagram under multicast tree cost of the present invention changes with network size;
Fig. 5 is that the present invention optimizes the multicast tree time cost with the performance comparison diagram under the network size variation;
To be multicast tree cost of the present invention count performance comparison diagram under the growth pattern with destination node to Fig. 6.
Embodiment
With reference to Fig. 1, specific implementation process of the present invention is as follows:
Step 1, initialization generates random network and the source node destination node is set, given operational factor and maximum allowable delay.This operational factor comprises initial population scale M, represent N the quantum bit position of alternative path collection of each destination node and the anglec of rotation δ of quantum rotation door.
With reference to Fig. 2, following steps obtain from a destination node to source node all satisfy the path of time delay condition:
2a. current will visit node be pressed in the stack of depositing the path, and its state is set to visit;
2b. if the current accessed node is a source node, the path of then preserving this arrival source node, and source node moved back stack;
2c. if present node is not a source node, and the node degree of present node is non-vanishing, then the time delay with each neighbor node compares with giving the fixed response time condition, be less than or equal to when fixed response time condition and its neighbor node are not in stack if add time delay behind its neighbor node, then return step 2a; Otherwise the present node node degree is subtracted 1;
2e. present node is moved back stack, and recovers its state and be visit;
Repeating step 2a tries to achieve the alternative path collection of all the other destination nodes to 2e, and the alternative path collection of each destination node of try to achieve is arranged by the cost ascending order.
Step 3 to the alternative path number that each destination node comprises, only selects wherein preceding 2
NBar, so just can use the alternative path collection of the point of each target joint of N quantum bit bit representation, when m destination node, the quantum bit individuality just has m * N bit, then according to the quantum evolution theory of computation, the path status matrix that obtains choosing each destination node is:
Wherein | α
Ij|
2+ | β
Ij|
2=1, (i=1,2 ..., m, j=1,2 ..., N) satisfy normalizing condition, and all a
Ij, β
IjAll be initialized as
Step 4 is to state matrix q
tThereby carry out quantum observation and decoding and calculate the multicast tree fitness function.
During quantum observation, with representing
The value that observation obtains, wherein x
Ij t(i=1,2 .., m, j=1,2 .. N) is a binary string that length is m * N.Concrete operations are as follows:
Produce the real number at random of [0,1], if it is greater than q
tIn | a
Ij t|
2, the corresponding positions value 0 of P (t) then, otherwise value 1 have so just obtained the string of binary characters of a m * N;
With the step-length is that N decodes to above-mentioned binary string, obtains the selected path of each destination node, for example works as N=3, t for the time,
What represent that second destination node choose is the 1st paths, and the rest may be inferred, 110 just expression choose the 7th paths of this purpose node, when certain destination node road total path number was 7,111,110 all represented to choose the 7th paths;
Through constituting a multicast tree, and calculate the fitness function Fitness=1/MultiTreeCost of this multicast tree by these routings, wherein MultiTreeCost is the multicast tree cost.
Step 5 is to representing the state matrix q of each destination node alternative path collection coding
t, carry out the quantum variation by the quantum revolving door.
Based on the basic framework of evolvement method, adopt quantum bit to represent individuality, design is at the quantum rotation door variation strategy of quantum individuality.In quantum theory, the migration between each state is that the transformation matrix of throughput cervical orifice of uterus is achieved, and equally also can characterize mutation operation in the quantum individuality with the anglec of rotation of quantum revolving door, and then
In variation, add the information of optimum individual, accelerate state matrix q
tConvergence.The quantum rotation door is defined as follows:
Here Δ θ
iBe the size of the anglec of rotation, θ
i=δ * s (α
i, β
i), δ state of a control matrix q wherein
tConvergence rate, s (α
i, β
i) be state of a control matrix q
tThe convergence direction.If the δ value is too little, convergence rate will slow down, if the δ value senior general might produce precocious phenomenon.
The present invention take a kind of general, adjust strategy with the irrelevant anglec of rotation of problem, use x
iAs the i position among the current common individual P (t); Use best
iAs the i position among the current optimum individual P (t), f (x) is a fitness function, s (α
iβ
i) control direction of rotation as shown in table 1:
Table 1 anglec of rotation Δ θ
iChoosing when δ=0.01 π is strategy then
As seen table 1 works as x
i=0, best
t=1, when f (x) is better than f (best), converge to a individuality with higher fitness for making current separating, should increase current separating and get 0 probability, promptly will make | α
i|
2Become big, according to quantum rotation door schematic diagram shown in Figure 3 as can be known, if α
iβ
i>0 when being illustrated in first and third quadrant, and θ should rotate to clockwise direction; If α
iβ
i<0 second, during four-quadrant, θ should be to rotation counterclockwise.
This moment state matrix q
T+1In i quantum bit position behind mutation operation be
Wherein
Be q
tIn i quantum bit position,
Step 6 is to the state matrix q of each destination node alternative path collection coding of back expression that makes a variation
T+1Carry out quantum observation, obtain one group of new binary string
And with step-length N P (t+1) is decoded and to obtain the path that each destination node is newly chosen, thereby calculate new multicast tree fitness function Fitness=1/MultiTreeCost, wherein MultiTreeCost is the multicast tree cost.If new multicast tree fitness function is better than the fitness function in the step 4, then adopt by the building mode of the structure multicast tree mode in the mode alternative steps 4 of these new routings through constituting multicast tree as optimum multicast tree, otherwise the optimum multicast tree building mode that continues to adopt the building mode conduct in the step 4 to search.
Step 7, judge whether to proceed to optimize according to end condition: just finish optimization method if satisfy end condition, establish final structure multicast tree mode, make except guaranteeing that all destination nodes can reach, also effectively reduce the network cost has also avoided by the overweight link of load as far as possible, information sends to different destination nodes with parallel mode along this branch, and this end condition is given maximum evolutionary generation.Otherwise, continue execution in step 5.
Effect of the present invention can further specify by following emulation content.
The emulation content: improved Waxman stochastic network model has been preserved in the generation of artificial network, and it has guaranteed to generate the connectedness of network.In this model, any two node u exist the probability of link to be between v
Wherein, (u v) is the Euclidean distance of node u to node v to d, L is a ultimate range between arbitrary node, and parameter alpha and β control produce the feature of network, and it is worth (0,1) between, the ratio on minor face and long limit in the parameter alpha Control Network, β are used for the average number of degrees of regulating networks node, and the coordinate space of the network of preservation is 4000km * 4000km, α is 0.26, β is 0.4, and the average number of degrees of node are 4, and time delay upper limit Δ is 100.
With GA and CS is comparison basis, preserves following performance index:
R wherein
CostAnd R
TimeBe respectively the average behavior that multicast is counted cost and find the solution multicast running time when counting, T represents experiment number, and R value represents that less than 1 average behavior of the present invention is better than GA and CS..
In the experiment, parameter of the present invention is set to: population scale M=10, the path of a destination node is chosen with N=2 quantum bit position.The parameter of existing CS is set to: population scale M=10, clone's scale 20, variation probability P
m=0.6.The parameter of existing GA is set to: population scale M=10, variation Probability p
m=0.03, crossover probability P
c=0.6.In CS and GA method, only select its preceding 10% for the alternative path collection of each destination node.For fairness relatively, all methods are an iteration 100 times all.All experimental results all are to obtain on the basis of independent operating 30 times.The environment of experiment operation is dominant frequency 2.33GHz Pentium IV and 2G internal memory, uses the C Plus Plus programming to realize.
Simulation result and analysis:
Fig. 4 is the performance comparison diagram under multicast tree cost of the present invention changes with network size, and wherein destination node accounts for 15% of network size.As can be seen from Figure 5, the performance of CS is better than GA, and the present invention is better than GA and CS, and at the increase R along with network size
CostStable performance, this explanation the present invention more is applicable to the optimization large scale network.
Fig. 5 is that the present invention optimizes the multicast tree time cost with the performance comparison diagram under the network size variation, and wherein destination node accounts for 15% of network size.As can be seen from Figure 6, the performance of GA is better than CS, and the present invention is better than GA and CS.Though at increase R along with network size
TimeThe trend that rises is slightly arranged, but generally with respect to GA and CS, the present invention still can finish optimization in the short period of time, satisfies the condition of in the network real-time being had relatively high expectations.
To be multicast tree cost of the present invention count performance comparison diagram under the growth pattern with destination node to Fig. 6, and wherein network size is 1000.From figure in 7 as can be seen, the performance of CS is better than GA, and the present invention is better than CS and GA.When having the multi-user to ask to communicate by letter simultaneously, the present invention also can satisfy the optimization to multicast tree.
In a word, the quantum evolution multicast routing optimization method that the present invention proposes can effectively solve multicast routing problem, has better ability of searching optimum than traditional heuristic, genetic method and Cologne system of selection, during optimization used time still less, and along with the growth of network size has good stable.
Claims (4)
1. the multicast routing optimization method based on quantum evolution comprises the steps:
(1) generates random network, and source node, destination node are set, given operational factor and maximum allowable delay in network;
(2) each given destination node is found the solution the alternative path collection that all satisfy the time delay condition, and arrange by ascending order;
(3) to preceding 2 of each destination node
NBar, the alternative path collection carries out quantum coding as follows, obtains a state matrix q
t, N ∈ [1,4] wherein:
3a) when m destination node, the length of calculating the individual coding of quantum bit is m * N bit;
3b) represent the state of the n paths that m destination node chosen successively with N bit wherein, the state matrix that obtains a quantum individuality is:
Here satisfy normalizing condition | α
Ij|
2+ | β
Ij|
2=1, (i=1,2 ..., m, j=1,2 ..., N), and all α
Ij, β
IjAll be initialized as
q
tRepresent one at t for the overlaying state of quantum bit individuality between 0 and 1;
(4) to representing the state matrix q of each destination node alternative path collection coding
tCarry out quantum observation, obtain one group of binary string;
(5) above-mentioned binary string is decoded, obtain the selected path of each destination node, through constituting a multicast tree, and calculate the fitness function of this multicast tree by these routings; Fitness=1/MultiTreeCost, wherein MultiTreeCost is the multicast tree cost;
(6) to representing the state matrix q of each destination node alternative path collection coding
t, carry out the quantum variation by the quantum revolving door;
(7) state matrix to each destination node alternative path collection coding after the expression variation carries out quantum observation, obtains one group of new binary string;
(8) new binary string is decoded, obtain the new routing of each destination node footpath, by these new routings through constituting a multicast tree, and calculate the fitness function of this tree, if new multicast tree fitness function is better than the fitness function in the step (5), then adopt the optimum multicast tree building mode that searches by the mode conduct of these new routings through constituting multicast tree, otherwise the optimum multicast tree building mode that continues to adopt the building mode conduct in the step (5) to search;
(9) judge whether to satisfy end condition,, finish to optimize and adopt the optimum multicast tree building mode that finally searches, information is sent to different destination nodes with parallel mode if satisfy end condition, otherwise, continued to carry out (5) step.
2. multicast routing optimization method according to claim 1, wherein the described given operational factor of step (1) comprises initial population scale M, represents N the quantum bit position of alternative path collection of each destination node and the anglec of rotation δ of quantum rotation door.
3. multicast routing optimization method according to claim 1, wherein step (2) is described finds the solution the alternative path collection that all satisfy the time delay condition to each given destination node, finds the solution as follows:
3a. with current will visit node be pressed in the stack of depositing the path, and its state is set to visit;
3b. if the current accessed node is a source node, the path of then preserving this arrival source node, and source node moved back stack;
3c. if present node is not a source node, and the node degree of present node is non-vanishing, then the time delay with each neighbor node compares with giving the fixed response time condition, be less than or equal to when fixed response time condition and its neighbor node are not in stack if add time delay behind its neighbor node, then return step 3a; Otherwise the present node node degree is subtracted 1;
3d. present node is moved back stack, and recovers its state and be visit;
3e. all the other destination nodes are carried out same operation according to 3a to 3d, and the alternative path of each destination node of try to achieve are arranged by the cost ascending order, finish solution procedure.
4. multicast routing optimization method according to claim 1, wherein step (5) is described decodes to binary string, obtains the selected path of each destination node, carries out as follows:
5a. by quantum observer state matrix q
t, the binary string that produces a group length and be m * N is:
5b. with N is that step-length is decoded to described binary string P (t), obtains the n bar road warp that i destination node chosen:
5c. the n bar road of choosing according to i destination node is through judging whether to count s greater than its total path, if n during greater than s, represents to choose is the s paths, otherwise choose be the n paths.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009100222896A CN101616074B (en) | 2009-04-30 | 2009-04-30 | Multicast routing optimization method based on quantum evolution |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009100222896A CN101616074B (en) | 2009-04-30 | 2009-04-30 | Multicast routing optimization method based on quantum evolution |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101616074A CN101616074A (en) | 2009-12-30 |
CN101616074B true CN101616074B (en) | 2011-08-24 |
Family
ID=41495493
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2009100222896A Expired - Fee Related CN101616074B (en) | 2009-04-30 | 2009-04-30 | Multicast routing optimization method based on quantum evolution |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101616074B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102158417A (en) * | 2011-05-19 | 2011-08-17 | 北京邮电大学 | Method and device for optimizing multi-constraint quality of service (QoS) routing selection |
CN103905319B (en) * | 2014-03-24 | 2017-02-01 | 中国电子科技集团公司第三十研究所 | Multiple-constraint multicast routing algorithm based on iteration coding |
CN104994021B (en) * | 2015-07-21 | 2018-06-15 | 三星电子(中国)研发中心 | Determine the method and device of optimal path |
CN106658639B (en) * | 2016-12-21 | 2020-05-12 | 天津理工大学 | QG-OLSR routing method based on quantum genetic strategy |
CN109450674B (en) * | 2018-10-25 | 2022-02-25 | 深圳市美格智联信息技术有限公司 | Detector DIS system man-machine interaction method based on multicast optimization |
CN109714261B (en) * | 2019-01-11 | 2020-12-25 | 东南大学 | Multicast routing method based on fidelity measurement in quantum communication network |
CN111669328B (en) * | 2020-07-02 | 2022-12-02 | 安徽省地震局 | Qos routing method based on quantum maximum minimum ant colony algorithm |
CN114358294B (en) * | 2022-02-22 | 2023-11-03 | 合肥本源量子计算科技有限责任公司 | Method, apparatus and storage medium for encoding raw data into quantum wire |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1378365A (en) * | 2002-05-10 | 2002-11-06 | 北京港湾网络有限公司 | IP group broadcast route repeating optimizing method |
CN101217500A (en) * | 2008-01-21 | 2008-07-09 | 重庆邮电大学 | A wireless self-organizing multicast routing energy improvement method based on MAODV protocol |
-
2009
- 2009-04-30 CN CN2009100222896A patent/CN101616074B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1378365A (en) * | 2002-05-10 | 2002-11-06 | 北京港湾网络有限公司 | IP group broadcast route repeating optimizing method |
CN101217500A (en) * | 2008-01-21 | 2008-07-09 | 重庆邮电大学 | A wireless self-organizing multicast routing energy improvement method based on MAODV protocol |
Non-Patent Citations (1)
Title |
---|
宋书强 等.基于量子粒子群算法的Qos组播路由的问题研究.《微电子学与计算机》.2009,第26卷(第4期),206-209. * |
Also Published As
Publication number | Publication date |
---|---|
CN101616074A (en) | 2009-12-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101616074B (en) | Multicast routing optimization method based on quantum evolution | |
CN113296845B (en) | Multi-cell task unloading algorithm based on deep reinforcement learning in edge computing environment | |
Patel et al. | A hybrid ACO/PSO based algorithm for QoS multicast routing problem | |
CN109818786B (en) | Method for optimally selecting distributed multi-resource combined path capable of sensing application of cloud data center | |
CN106297774B (en) | A kind of the distributed parallel training method and system of neural network acoustic model | |
CN107220731A (en) | A kind of logistics distribution paths planning method | |
CN111325356A (en) | Neural network search distributed training system and training method based on evolutionary computation | |
CN114143264B (en) | Flow scheduling method based on reinforcement learning under SRv network | |
Zhang et al. | The research of genetic ant colony algorithm and its application | |
CN112235385B (en) | Offshore edge computing low-overhead cloud edge intelligent cooperative ally member discovery method | |
CN107911300B (en) | Multicast routing optimization method based on whale algorithm and application of multicast routing optimization method on Spark platform | |
CN107094112A (en) | Bandwidth constraint multicast routing optimization method based on drosophila optimized algorithm | |
CN111885493B (en) | Micro-cloud deployment method based on improved cuckoo search algorithm | |
CN103824127B (en) | Service self-adaptive combinatorial optimization method under cloud computing environment | |
CN101447936B (en) | Multicast routing method based on particle swarm algorithm | |
Xu et al. | Living with artificial intelligence: A paradigm shift toward future network traffic control | |
Zhao et al. | Reinforcement learning for resource mapping in 5G network slicing | |
CN109800911B (en) | Unified navigation method for delivery paths of multiple couriers | |
CN114154685A (en) | Electric energy data scheduling method in smart power grid | |
CN102158413B (en) | Multi-agent multicast routing method based on adjacent immune clonal selection | |
CN101741749A (en) | Method for optimizing multi-object multicast routing based on immune clone | |
CN114118444B (en) | Method for reducing equipment idle running time in federal learning by using heuristic algorithm | |
Sun et al. | QPSO-based QoS multicast routing algorithm | |
Vijayalakshmi et al. | Artificial immune based hybrid GA for QoS based multicast routing in large scale networks (AISMR) | |
Peleteiro et al. | How coalitions enhance cooperation in the ipd over complex networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20110824 Termination date: 20170430 |