CN111770020B - Method for network perception service combination algorithm based on optimal path selection - Google Patents
Method for network perception service combination algorithm based on optimal path selection Download PDFInfo
- Publication number
- CN111770020B CN111770020B CN202010586754.5A CN202010586754A CN111770020B CN 111770020 B CN111770020 B CN 111770020B CN 202010586754 A CN202010586754 A CN 202010586754A CN 111770020 B CN111770020 B CN 111770020B
- Authority
- CN
- China
- Prior art keywords
- service
- path
- algorithm
- node
- network
- 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.)
- Active
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 121
- 238000000034 method Methods 0.000 title claims description 30
- 230000008447 perception Effects 0.000 title description 4
- 238000010586 diagram Methods 0.000 claims abstract description 6
- 239000000203 mixture Substances 0.000 claims description 32
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 230000004044 response Effects 0.000 claims description 7
- 101100391182 Dictyostelium discoideum forI gene Proteins 0.000 claims description 6
- 239000002243 precursor Substances 0.000 claims description 2
- 230000008901 benefit Effects 0.000 abstract description 8
- 230000008569 process Effects 0.000 description 16
- 238000012360 testing method Methods 0.000 description 7
- 238000005457 optimization Methods 0.000 description 5
- 239000002131 composite material Substances 0.000 description 4
- 101000742062 Bos taurus Protein phosphatase 1G Proteins 0.000 description 3
- 125000002015 acyclic group Chemical group 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 235000012054 meals Nutrition 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- 235000008694 Humulus lupulus Nutrition 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000005469 granulation Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 208000011580 syndromic disease Diseases 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 239000008187 granular material Substances 0.000 description 1
- 230000003179 granulation Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/302—Route determination based on requested QoS
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention belongs to the technical field of network service, in particular to a network service systemFirstly, inputting a graph G (V, E, s, t, omega, W, C, C, H, K) and a parameter epsilon; then, according to the node constraint of the service network graph path in the step one, the topological nodes which do not meet the service capability are measured in advance; recalculating new weight of each edge according to the simplified service network diagram in the step two, and setting a search space parameter delta; initializing a (K-1) dimensional path weight array; calculating an optimal path by using dynamic programming, and recording path nodes; and finally, searching the optimal service combination path meeting the user requirement. The OSP algorithm of the invention carries out the value taking on a given epsilon, the OSP algorithm can more quickly find the same service combination path as the ADAPT algorithm, compared with the ADAPT algorithm, the OSP algorithm has more obvious advantages in the aspect of large AET performance, and meanwhile, the OSP algorithm carries out the value taking on different epsilon and W2And the value is taken, a service combination path meeting the QoS requirement of a user is found more quickly, and the OSP algorithm has strong stability.
Description
Technical Field
The invention belongs to the technical field of network services, and particularly relates to a method for a network-aware service combination algorithm based on optimal path selection.
Background
With the continuous development of modern information technologies such as cloud computing, internet of things, mobile internet, big data and the like, the requirements of users (especially mobile phone users) are increasing day by day, a single service cannot meet the complex requirements of the users, and the service combination technology integrates the existing distributed services into a combined service meeting the requirements of the users. Under the background that the personalized demand is continuously expanding, with the release of a large amount of mobile phone APP software, how to combine different services and ensure the service quality of a global network becomes a problem to be solved urgently in the present society.
Web services technology is closely related to our daily lives. When a user opens the mobile phone APP to select a travel plan, each service class (also called task or abstract service, such as booking a train ticket, booking a high-speed railway meal booking service, booking a car taking service, booking a hotel) has a plurality of combined additional services. They can fulfill the function of the class of service they are in, but they have different qualities of service. Therefore, one selects an optimal service composition scheme according to the composition service it provides. This service form of cloud computing is highly dependent on network service, i.e., network performance between the cloud and the user side. When a user selects a certain service, the service quality of the user often changes according to the network service quality, that is, the selection of one candidate service affects the service quality of another candidate service. In the service combination application, because the network resource allocation adopts a first-come-first-serve mode, when multiple users select corresponding services at the same time, once the computing resources are insufficient, network congestion occurs, and the performance of the network service quality is greatly reduced. Meanwhile, different kinds of candidate services are forwarded according to the routing principle in the original path, and the service quality on a network transmission layer is difficult to guarantee. Therefore, the QoS attributes of the cloud service and the network service are considered at the same time, the service combination problem in an integrated environment is solved, an optimal service combination algorithm is provided, point-to-point service quality guarantee is provided for users with different requirements, and the method has important theoretical value and wide application prospect.
A lot of research has also been done by many scholars on the aspect of service portfolio optimization. Firstly, constructing a Task-granulation layered service combination model and analyzing the computational complexity of the model; secondly, theoretically analyzing the feasibility of the QoS attribute in the task granulation process according to the existing QoS attribute calculation mode; zelina introduces a social relationship theory in solving the problem of a mass of online-to-offline (O2O) service environment, considers the improvement of the cooperation efficiency among offline service providers in the online service combination stage, and simultaneously optimizes the execution efficiency of an algorithm. Firstly, establishing a social relationship network model capable of reflecting the cooperation efficiency among offline service providers; secondly, a service filtering stage is added before an online optimization stage, a Skyline filtering method for social relationship expansion is provided, and the execution efficiency of the combined service in a massive service environment is improved; and finally, in the service combination optimization stage, adding a local search operator aiming at the cooperation efficiency in the multi-target genetic algorithm. Aiming at the problem of service combination, the Chua Jiangan and the like establish a service combination optimization model taking time, practicability, innovation, reliability and the like as optimization targets by using a service combination strategy; meanwhile, the path search space is preprocessed through the clustering algorithm and the association rule of the combined service, so that the efficiency of the search space is greatly improved, the knowledge service resources can be accurately positioned and matched with the service combination in a short time, and the performance of the service combination is improved to a great extent.
Funda Ergun et al propose an approximation algorithm (ADAPT) with excellent performance for the dual-metric QoS routing problem, and introduce an approximation test process of polynomial time by selecting an optimal upper and lower bound to continuously reduce the distance between the upper and lower bound, convert the path search problem into an interval search problem, and quickly find the optimal solution within the continuously reduced range. Guolang Xue et al propose a Peseudo MCPP algorithm, and provide a path search space constructed by using path hops, so that under the condition of satisfying the first-dimension path weight constraint D, the path weights of other dimensions also satisfy a given beam C. Since the algorithm proposed by Funda Ergun is very time consuming, the latter algorithm cannot inherit known results, and it is difficult to obtain a better quality solution.
Disclosure of Invention
On the basis of comprehensively considering the characteristics of the network service, the invention combines the two algorithm ideas to provide an optimal path selection algorithm (OPS) for the service so as to solve the problem of QoS-aware network service combination, improve the level of network service quality and improve the quality of user experience.
The invention is realized in this way, a method of network perception service combination algorithm based on optimal path selection, comprising the following steps:
the method comprises the following steps: inputting a graph G (V, E, s, t, omega, W, C, C, H, K) and a parameter epsilon;
step two: according to the node constraint of the service network graph path in the step one, topology nodes which do not meet the service capability are measured in advance;
step three: recalculating new weight of each edge according to the simplified service network diagram in the step two, and setting a search space parameter delta;
step four: initializing a (K-1) dimension path weight array;
step five: calculating an optimal path by using dynamic programming, and recording path nodes;
step six: finding the optimal service combination path meeting the user requirement,
wherein: assume a service composition scheme, such as a path p in a service network, if for
Is provided with cv≥ChAnd ωk(p)≤WkWhere v is e ShH is 1 ≦ H, K is 1 ≦ K, then path p is a feasible service composition scheme, given that for a given graph G (V, E) all feasible service composition schemes are denoted as { p ≦ KfThen for any one of the service composition schemes
All have a minimum value ηi∈(0,1]So that
And for
Is provided with cv≥Ch，1≤h≤H，
And carrying out topology structure conversion on the data: sa and Sb are kept unchanged during conversion, and after Sa is executed, p is used as p1,p2,L,pnProbabilistic execution service S1,S2,L,SnThen Sb is performed. t is ti,ci,ri,thiRespectively represented response time, cost, reliability and throughput, Fi(x),Pi(x),Pi(x),Pi(x) Respectively representing the corresponding probability response time t and the probability P thereof, the cost c and the probability P thereof, the reliability r and the probability P thereof, the throughput th' and the probability P thereof respectively represent the selection structureThe value of the converted service S1 n.
Preferably, in step two: the condition that the service capability of the node upsilon cannot meet the service requirement is judged to be cυ<Ch。
Preferably, in step three: the original service network graph is directly transformed into a simpler graph after the nodes in the step two are removed, and the simpler graph is used for solving a feasible solution of the QoS-aware service composition problem, wherein: z is a radical ofi{ (α, β) | α ∈ V, β ∈ E '}, where E' denotes a new set of weights for the association node.
Preferably, step four: wherein the array dυ[δ2,L,δK]Recording the path p from the source node s to any intermediate node upsilon, and recording the minimum weight (omega) of the first dimension parameter of the path p1) Furthermore, the path p satisfies ωk(p)≤δk，2≤k≤K；
Array pυ[δ2,L,δK]Recording a precursor node of a node upsilon on a path p, wherein the path p satisfies omega1(p)＝dυ[δ2,L,δK]And ω isk(p)≤δk，2≤k≤K。
Preferably, step five: when u-upsilon is an edge in the graph, a first-dimension minimum weight path p from a node s to a node upsilon must pass through some intermediate nodes u, and when a service optimal path selection algorithm (OSP) searches the edges between all possible intermediate nodes u and the node upsilon, the minimum first-dimension path weight d is obtainedυ[δ2,L,δK]；
If the weight d of the currently recorded first dimension pathυ[δ2,L,δK]Greater than the previously recorded value dυ[δ2,L,δj-1,L,δK]Then inherit the previous first dimension minimum path weight.
Preferably, step six: if the path P satisfies ωk(p)≤WkAnd if K is more than or equal to 2 and less than or equal to K, returning the feasible path p by the algorithm, otherwise, returning a prompt of no feasible path, and terminating the program.
Compared with the prior art, the invention has the beneficial effects that:
according to the inventionThe service optimal path selection algorithm (OSP) carries out value taking on a given epsilon, the service optimal path selection algorithm (OSP) can find a service combination path which is the same as the service combination path for running the ADAPT algorithm more quickly, namely the AET value of the service optimal path selection algorithm (OSP) is smaller, the advantage of the service combination path is more obvious along with the increase of the service number, the algorithm adopted by the ADAPT algorithm is slower in process because the upper and lower boundaries of the path are gradually compressed, and the finally constructed search space is larger than the search space constructed by depending on the hop number, so compared with the ADAPT algorithm, the service optimal path selection algorithm (OSP) has more obvious advantage in large AET performance, and meanwhile, different epsilon and W are evaluated by the service optimal path selection algorithm (OSP)2Value taking, the service optimal path selection algorithm (OSP) is superior to the ADAPT algorithm, namely the service combination path meeting the user QoS requirement can be found more quickly, along with the increase of the value of the algorithm parameter epsilon, the AET value of the service optimal path selection algorithm (OSP) is more stable, and the AET value of the ADAPT algorithm is gradually reduced, which shows that when the service combination path meeting the user requirement exists at a certain specific position in the search space, the ADAPT algorithm is sensitive to the value of the parameter epsilon and is influenced by W2The service optimal path selection algorithm (OSP) has better use effect, higher efficiency and practicability.
Drawings
FIG. 1 shows an example of a combined service of a certain travel network service;
FIG. 2 is a selected structural simplification;
FIG. 3 is a perceptual service composition model;
FIG. 4 is a service optimal Path selection Algorithm (OSP) design flow;
FIG. 5 is a graph of different service number AET values;
FIG. 6 shows the values of ε for 500 and 1000 service numbers;
FIG. 7 shows ε and W at the same number of services (500 and 1000)2The experimental results of (1);
FIG. 8 shows W for 500 and 1000 service numbers2A PWD value of;
FIG. 9 is a diagram of different service number PWD values;
FIG. 10 shows a comparison of service optimal Path selection algorithm (OSP) performance.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example (b):
as shown in fig. 1, for a certain travel case, each service class is represented by { (S1, S2, S3, …, St) }, and we can divide into 4 service instances according to the corresponding body of the composite service, that is: taxi taking platform service, online ticket buying, online meal ordering and hotel reservation. The task granule (abstract service) is represented as { (S1, S2, S3, S4) }. If each task is assumed to have related candidate services, the function of the service class where the task is located can be realized, but the tasks have different service qualities. Therefore, one would select the optimal service composition scheme based on the quality of service of the candidate service.
The service combination can fully utilize the existing services at present and has the advantages of real-time, quick and flexible establishment of distributed loose coupling. When a network user requests a service from the server, the server calls a service combination process and selects a series of atomic service components to construct a combined service with complex functions so as to meet the personalized requirements of the network user. The service composition path in this patent includes a plurality of individual services combined, i.e., a composite service (also called multi-target service composition, multi-task composition service, or comprehensive service).
Define 1 a feasible service composition scheme. Assume a service composition scheme, such as a path p in a service network, if for
Is provided with cv≥ChAnd ωk(p)≤WkWhere v is e ShH is more than or equal to 1 and less than or equal to H, K is more than or equal to 1 and less than or equal to K, and then the path p is a feasible service combination scheme.
For a given graph G (V, E), assume that all possible service composition schemes are denoted as { p }f}, thenFor any one of the service composition schemes
All have a minimum value ηi∈(0,1]So that
And for
Is provided with cv≥Ch，1≤h≤H。
When a user selects a candidate service in each task particle, the candidate service directory selects an optimal combined path meeting the user requirements from a large number of candidate service combined path flows according to the QoS requirements of the user, because each candidate service combined path includes a plurality of topological structures, most of the existing service selection algorithms cannot directly process the plurality of structures, and therefore before service selection, structure conversion needs to be performed on the topology, and fig. 2 is a process for converting the selection structure.
Sa and Sb remained unchanged during this conversion as shown in fig. 2. After Sa is executed, p is used1,p2,L,pnProbabilistic execution service S1,S2,L,SnThen Sb is performed. t is ti,ci,ri,thiRespectively represented response time, cost, reliability and throughput, Fi(x),Pi(x),Pi(x),Pi(x) Respectively representing the corresponding probability response time t and the probability P thereof, the cost c and the probability P thereof, the reliability r and the probability P thereof, and the throughput th' and the probability P thereof respectively represent the values of the service S1n after the structure is selected for conversion. The calculation is as follows:
in the service composition model, a single service generally cannot meet the user requirements, and the requirements are to combine a plurality of services into a workflow in sequence to complete the user request. In the composite service selection process, there are typically multiple candidate service providers for a service. As shown in fig. 3, each service St(1. ltoreq. T. ltoreq.T) with a plurality of atomic service modules Stf(1 ≦ i ≦ T, 1 ≦ F ≦ F) as a candidate component for the service composition. The perceptual service composition model maps the QoS parameters of the network service and the cloud service together on the edge between every two services. For each service request, the service composition process will find a path p from the service entry s to the service exit t. If the path p intersects all the selected service components and meets the user QoS requirements, it will be returned to the user as a composite service (or service composition scheme).
In the case shown in fig. 1, when four tasks of the taxi taking platform service, the online ticket, the online meal ordering, and the hotel reservation are cloud services, the corresponding network quality can be set as the relevance service.
An association criterion is said to exist between candidate services in a service class when the quality of service of the candidate services is dependent on the quality of service of candidate services in another service class. Obviously, in an association criterion, if the quality of service of one candidate service changes, the quality of service of another candidate service associated necessarily changes. When a plurality of candidate services are involved in one relevance criterion, accordingly, a change in the quality of service of some candidate services necessarily causes a change in the quality of service of some other candidate services. From the mathematical definition, zi{ (α, β) | α ∈ V, β ∈ E '}, where E' denotes a new set of weights for the association node. Specifically,. alpha.epsilon. { v'1,v′2,L,v′IDenotes I correlation nodes, correspondingly, β ∈ { ω'1(e),ω′2(e),L,ω′I(e) Denotes I new weights.
Strict mathematical definitions are given according to the process of establishing the association matcher in the literature and the symbols and corresponding meanings required in the establishment process. In consideration of the association between tasks, some tasks that have already been completed may affect the correspondence between the unexecuted tasks and the specific services. In FIG. 3, service S1、S2、S3、S4As in the candidate service (S)1f，S2f，…,Stf) The completed task grains are all a group of task grains with specific association. When the service selects a task, firstly, the task particles with the optimal service quality and associated with the service need to be considered. And when the service with the association relation can not meet the requirement of the user, re-selecting the task particles. The steps of the association re-selection algorithm are as follows:
1) updating parameters of the service composition model: the set of tasks that have been completed is Stf＝{S1f，S2f，…,StfAnd E (S) is corresponding to the service for completing the tasktf)＝{S1f，S2f，…,StfThen S is processedtfAnd E (S)tf) Parameter x of corresponding relation between1，x2，…，xuIs set to 1. Set E (S)tf) The price and the response time of the contained services in the actual execution process are substituted into the service combination model, and the corresponding reliability of the services in the combination model is set to be 1.
2) Candidate service selection: set a set of associated tasks as S1f＝{S11，S12，…,S1fS is the corresponding candidate service set with QoS incidence relation2f＝{S21，S22，…,S2f}. Task set S1fE S is a set of tasks that have been completed, corresponding to E (S)tf) Is E (S)tf)＝{S1f，S2f，…,StfAnd then for any set of associated services. If satisfying the bar Stf∈E(Stf) Then the association is made.
3) Correlating the results obtained in the two steps by using a correlation criterion, judging by using the parameters of the service combination model, and correlating the tasks when the parameters are 1; when the parameter is 0, the task association fails.
The QOS perception-based service combination algorithm comprises the following steps:
it is fully feasible to find an approximately optimal solution with less resource consumption within the range that the actual scene can bear. Designing a network service combination approximation algorithm for solving the problem according to the approximation algorithm theory is a very reliable and effective method. Therefore, in combination with the idea of approximate algorithm, the patent provides a service optimal path selection algorithm (OSP) to obtain a network service combination scheme satisfying the QoS requirements of users, and the flow chart of the algorithm is shown in fig. 4.
Pseudo-code of the network-aware service composition service optimal path selection algorithm (OSP) is shown in table 1.
TABLE 1
The service optimal path selection algorithm (OSP) mainly comprises the following four specific steps:
(1) the first step (line 1), topology pruning is based on node constraints. In the service network, cυ<ChMeaning that the service capability of node v cannot meet the service requirement, the node may be excluded from the path search space in advance.
(2) Second step (line 2), calculate the new weight of each edge, set parameter Δ. At this time, the original service network diagram is directly converted into a simpler diagram, and the purpose of conveniently solving a feasible solution of the QoS-aware service combination problem is achieved.
(3) And thirdly (lines 3-19), initializing a (K-1) dimensional array, running a dynamic planning process and searching a path from the service entrance node s to the service exit node t.
R array dυ[δ2,L,δK]Recording the path p from the source node s to any intermediate node upsilon, and recording the minimum weight (omega) of the first dimension parameter of the path p1). In addition, the path p satisfies ωk(p)≤δk，2≤k≤K。
② array pυ[δ2,L,δK]The predecessor node of node v on path p is recorded. The path p satisfies ω1(p)＝dυ[δ2,L,δK]And ω isk(p)≤δk，2≤k≤K。
Row 10-13: when u-upsilon is an edge in the graph, a first-dimension minimum weight path p from a node s to the node upsilon must pass through some intermediate nodes u, so that the service optimal path selection algorithm (OSP) provided by the patent obtains the minimum first-dimension path weight d by searching the edges between all possible intermediate nodes u and the node upsilonυ[δ2,L,δK]。
(4) And step four (lines 20-26), searching whether the path obtained by the calculation result of the step three is feasible or not and whether the constraint condition provided by the user is met or not. If the path P satisfies ωk(p)≤WkAnd if K is more than or equal to 2 and less than or equal to K, returning the feasible path p by the algorithm, otherwise, returning a prompt of no feasible path, and terminating the program.
And (3) algorithm analysis:
theorem 1 the time complexity of the serving optimal Path selection Algorithm (OSP) is
(H E shows that it takes O (n + m) to delete topology, O (m) to calculate the edge weight, and O (m) to calculate the path
And fourthly, testing whether the obtained path is feasible or not, wherein the time is O (K). Thus, the time complexity of the service optimal path selection algorithm (OSP) in the worst caseIs composed of
After the syndrome is confirmed.
And (3) proving that: for the optimal path poptHas omegak(popt)≤ηopt·WkK is 2. ltoreq. k.ltoreq.K, i.e
Because of the fact that
And is
Then
From dυ[δ2,L,δK]The process of definition and dynamic planning of
Can obtain the product
Because of the fact that
Then
Since path p has H-1 hops, therefore
Namely, it is
ωk(p)≤(1+ρ)·η·Wk (23)
Wherein,
after the syndrome is confirmed.
In order to verify the performance of the service optimal path selection algorithm (OSP), the patent performs different comparison tests on the algorithm under different environments.
An MATLAB10 simulation software platform is adopted in the simulation, and a hardware platform is an InterI 7 dual-core processor and a PC of 16G DDR 4. In the process of testing the performance of the algorithm, the patent utilizes MATLAB10 to generate two groups of directed acyclic graphs as data sets. The first set of data sets includes 10 directed acyclic service networks with a service number scale of 100-. The first QoS parameter takes a value randomly in the range of (2, 10), and the second QoS parameter takes a value randomly in the range of (10-8, 10-5).
The second set of data sets comprises 5 directed acyclic service networks with a service number size of 20-100, wherein each service network also has 3 service classes (H-3), and each directed edge between service numbers has three QoS parameters, i.e., K-3. The first QoS parameter takes a value randomly in a range of (2, 10), the second QoS parameter takes a value randomly in a range of (10-8, 10-5), and the third QoS parameter takes a value randomly in a range of (0, 0.01).
In order to evaluate the performance effectiveness of the OSP, the method adopts two evaluation indexes to reflect the algorithm solving Time and solving quality, namely Average Execution Time (AET) and Path Weight Distance (PWD), so as to achieve the purpose of testing the comprehensive performance of the algorithm.
AET is mainly used for evaluating indexes of the algorithm in terms of network delay, and when the delay required by a combined path spent by a network user is smaller, the performance of the solution time is better. The distance between the service combination path weight and the network user path constraint obtained by PWD under different operation algebra mainly is
For the end user, the evaluation index reflects the QoS guarantee level of the service combination path obtained by the algorithm. Obviously, the larger the PWD value, the better the quality of the service composition path obtained by the algorithm.
Based on the foregoing analysis, the present patent introduces an approximation parameter ε into the service-optimal path selection algorithm (OSP). In the test, different epsilon values are respectively set to analyze the influence of the epsilon values on the QoS, and the epsilon values are respectively 0.01, 0.02 and 0.03. For a user end-to-end QoS request, assume W1＝50，W2∈[2.5×10-5,5×10-5]To ensure that the algorithm can find a feasible service combination in all service networks. In the experiment, the algorithm is operated for 1000 times of iteration under different values of epsilon, a union set of 20 operation results is obtained, and the average time value of the service combination path is obtained. FIG. 5 shows the service optimal Path selection Algorithm (OSP) at different service number scales [100-]And then, obtaining the average execution time required by the service combination path meeting the user requirement. AE when the number of services increases in sizeThe value of T increases, i.e. the trend in larger scale service networks, takes more time to find the same path. Along with the reduction of the parameter epsilon value, the time spent on searching and calculating the path is increased by the AET value, which is mainly because the smaller the value of the algorithm parameter epsilon is, the larger the path searching space is. Thus, the parameter ε reflects the approximation of the algorithm. The smaller the value of epsilon, the higher the approximation degree, and the larger the search space of the algorithm. Obviously, the algorithm needs to spend more time searching for the service combination path meeting the user requirement. W2(x10-5)。
FIG. 6 shows the trend of the ATE value of the algorithm as the value of ε varies. For a given epsilon value, in the range of service number [100,1000], the search space of the service optimal path selection algorithm (OSP) depends on the hop number of the service combination path, the path hop value is small and can be obtained in advance, and the search space is easily, quickly and effectively constructed. The service optimal path selection algorithm (OSP) can find a service combination path faster, the average execution time AET value of the algorithm is smaller, and the method is more prominent particularly under the condition of larger service number.
Fig. 7(a) shows the value of AET for the case where the number of services is 500 and 1000, and the values of e are different. It can be seen from the figure that different values of epsilon, the service optimum path selection algorithm (OSP) can meet the service combination path required by the user QoS faster. In addition, as the value of the algorithm epsilon increases, the OSF algorithm also exhibits more stable AET performance, that is, when a service combination path meeting the user requirement exists at a certain specific position in the search space, the AET value is more stable. The parameter epsilon constructs a search space in the hop count of the service combination path, and the parameter epsilon plays a role in fine tuning the search space. FIG. 7(b) shows different constraint values (W) for the number of services 500 and 10002) The value of AET in the case of a value. The experimental result shows that under different constraint conditions, the AET value of the service optimal path selection algorithm (OSP) is not influenced by user constraint, and the trend of the AET value in the algorithm is stable as the user constraint value is increased, and the AET value is always kept in a stable state.
FIG. 8 shows the service optimal Path selection Algorithm (OSP) at different constraint values (W)2) PWD value of the case. As the user constraints are gradually relaxed, the path distance searched by the service optimal path selection algorithm (OSP) will decrease, i.e., the PWD value decreases. That is, if the user's demand is relaxed, the service optimal path selection algorithm (OSP) will find a path with coarser quality, only to meet the user's requirements. It can be seen that the service optimal path selection algorithm (OSP) can find a better quality service combination path as the number of services increases.
FIG. 9 shows the PWD values for the service optimal Path selection Algorithm (OSP) at different network service numbers. As can be seen from the figure, as the number of network services increases, the service optimum path selection algorithm (OSP) can find a better quality service combination path. This is because there is a better service combination selection path in a larger service network. This shows that the service optimal path selection algorithm (OSP) can stably and reliably find a service combination path satisfying end-to-end QoS constraints under different service network numbers according to the user requirements.
To verify the performance of the service-optimal path selection algorithm (OSP), we compared the literature algorithm with the service-optimal path selection algorithm (OSP). Fig. 10(a) shows the AET contrast values for the two algorithms at different network service numbers, the same epsilon value. Test results show that for a given value of epsilon, the service-optimized path selection algorithm (OSP) can find the same service combination path as running the ADAPT algorithm faster, i.e. the AET value of the service-optimized path selection algorithm (OSP) is smaller. Its advantages become more apparent as the number of services increases. The algorithm adopted by the ADAPT algorithm is slow in process because the upper and lower boundaries of the path are gradually compressed, and the finally constructed search space is larger than the search space constructed by depending on the hop count. Therefore, the advantages of service-optimal path selection algorithm (OSP) large AET performance are more apparent compared to the ADAPT algorithm.
FIGS. 10(b) and 10(c) show the same service number (500 and 1000), different values of epsilon and different constraint values (W) for the two algorithms respectively2(10-5) AET contrast value of). For different epsilon and W2The value taking and the service optimal path selection algorithm (OSP) are all superior to the ADAPT algorithm, so that the satisfied use can be found more quicklyService composition paths for user QoS requirements. With the increase of the value of the algorithm parameter epsilon, the AET value of the service optimal path selection algorithm (OSP) is stable, and the AET value of the ADAPT algorithm is gradually reduced. This shows that when a service composition path meeting the user's requirements exists at a specific location in the search space, the ADAPT algorithm is sensitive to the parameter epsilon and is affected by W2And the stability of the service optimal path selection algorithm (OSP) is strong.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (3)
1. A method of network-aware service combination algorithm based on optimal path selection is characterized by comprising the following steps:
the method comprises the following steps: inputting a graph G (V, E, s, t, omega, W, C, C, H, K) and a parameter epsilon;
step two: according to the node constraint of the service network graph path in the step one, topology nodes which do not meet the service capability are measured in advance; wherein the condition that the service capability of the node upsilon cannot meet the service requirement is judged to be cυ<Ch；
Step three: recalculating new weight of each edge according to the simplified service network diagram in the step two, and setting a search space parameter delta;
step four: initializing a (K-1) dimension path weight array;
step five: calculating an optimal path by using dynamic programming, and recording path nodes; when u-upsilon is an edge in the graph, a first-dimension minimum weight path p from a node s to a node upsilon must pass through some intermediate nodes u, and when a service optimal path selection algorithm (OSP) searches the edges between all possible intermediate nodes u and the node upsilon, the minimum first-dimension path weight d is obtainedυ[δ2,L,δK](ii) a If the weight d of the currently recorded first dimension pathυ[δ2,L,δK]Greater than the previously recorded value dυ[δ2,L,δj-1,L,δK]Inheriting the previous first-dimension minimum path weight;
step six: finding the optimal service combination path meeting the user requirement,
wherein: a service composition scheme, a path p in a service network, if for
Is provided with cv≥ChAnd ωk(p)≤WkWhere v is e ShH is 1 ≦ H, K is 1 ≦ K, then path p is a feasible service composition scheme, given that for a given graph G (V, E) all feasible service composition schemes are denoted as { p ≦ KfThen for any one of the service composition schemes
All have a minimum value ηi∈(0,1]So that
K is 1. ltoreq. K, and for
Is provided with cv≥Ch，1≤h≤H，
And carrying out topology structure conversion on the data: sa and Sb are kept unchanged during conversion, and after Sa is executed, p is used as p1,p2,L,pnProbabilistic execution service S1,S2,L,SnThen Sb is executed; t is ti,ci,ri,thiRespectively represented response time, cost, reliability and throughput, Fi(x) Probability density function, P, representing the probability response time ti1(x) Density function, P, representing cost ci2(x) Density function, P, representing reliability ri3(x) A density function representing the throughput th', P representing the value of the service S1n after the sequential structure conversion; if the path P satisfies ωk(p)≤WkK is more than or equal to 2 and less than or equal to K, the algorithm returns to be feasibleAnd if not, returning no feasible path prompt and terminating the program.
2. The method of claim 1, wherein the step three is as follows: the original service network graph is directly transformed into a simpler graph after the nodes in the step two are removed, and the simpler graph is used for solving a feasible solution of the QoS-aware service composition problem, wherein: z is a radical ofi{ (α, β) | α ∈ V, β ∈ E '}, where E' denotes a new set of weights for the association node.
3. The method for network-aware service combination algorithm based on optimal path selection as claimed in claim 1, wherein the fourth step: wherein the array dυ[δ2,L,δK]Recording the path p from the source node s to any intermediate node upsilon, and recording the minimum weight (omega) of the first dimension parameter of the path p1) The path p satisfies ωk(p)≤δk，2≤k≤K；
Array pυ[δ2,L,δK]Recording a precursor node of a node upsilon on a path p, wherein the path p satisfies omega1(p)＝dυ[δ2,L,δK]And ω isk(p)≤δk，2≤k≤K。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010586754.5A CN111770020B (en) | 2020-06-24 | 2020-06-24 | Method for network perception service combination algorithm based on optimal path selection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010586754.5A CN111770020B (en) | 2020-06-24 | 2020-06-24 | Method for network perception service combination algorithm based on optimal path selection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111770020A CN111770020A (en) | 2020-10-13 |
| CN111770020B true CN111770020B (en) | 2022-04-12 |
Family
ID=72722291
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010586754.5A Active CN111770020B (en) | 2020-06-24 | 2020-06-24 | Method for network perception service combination algorithm based on optimal path selection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111770020B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112953760B (en) * | 2021-01-27 | 2022-05-27 | 华侨大学 | Low-cost large-scale personalized service customization method facing service value |
| CN114239900A (en) * | 2021-10-13 | 2022-03-25 | 三明学院 | Optimal path selection method, device, equipment and readable storage medium |
| CN116738246B (en) * | 2023-06-12 | 2023-12-26 | 烟台大学 | Combined service dynamic reconstruction method and system for service demand change |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101719932A (en) * | 2009-11-20 | 2010-06-02 | 中国科学院计算技术研究所 | System and method for automatic service combination |
| 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 |
| KR20130078037A (en) * | 2011-12-30 | 2013-07-10 | 건국대학교 산학협력단 | Qos-aware web service composition method using on-the-fly learning-based search |
| KR20130078041A (en) * | 2011-12-30 | 2013-07-10 | 건국대학교 산학협력단 | Large scale qos-aware web service composition method using efficient anytime algorithm |
| CN109167833A (en) * | 2018-09-05 | 2019-01-08 | 河海大学 | A kind of expansible QoS perception combined method based on figure |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8982709B2 (en) * | 2007-01-31 | 2015-03-17 | Hewlett-Packard Development Company, L.P. | Selecting service nodes for an end-to-end service path from a reduced search space |
| CN101471868A (en) * | 2007-12-27 | 2009-07-01 | 华为技术有限公司 | Route selection method and network system, route calculation module |
-
2020
- 2020-06-24 CN CN202010586754.5A patent/CN111770020B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101719932A (en) * | 2009-11-20 | 2010-06-02 | 中国科学院计算技术研究所 | System and method for automatic service combination |
| 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 |
| KR20130078037A (en) * | 2011-12-30 | 2013-07-10 | 건국대학교 산학협력단 | Qos-aware web service composition method using on-the-fly learning-based search |
| KR20130078041A (en) * | 2011-12-30 | 2013-07-10 | 건국대학교 산학협력단 | Large scale qos-aware web service composition method using efficient anytime algorithm |
| CN109167833A (en) * | 2018-09-05 | 2019-01-08 | 河海大学 | A kind of expansible QoS perception combined method based on figure |
Non-Patent Citations (1)
| Title |
|---|
| Lecture 02:Faster Approximations for QoS Routing;jian69309;《豆丁》;20130508;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111770020A (en) | 2020-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111770020B (en) | Method for network perception service combination algorithm based on optimal path selection | |
| Elgamal et al. | Costless: Optimizing cost of serverless computing through function fusion and placement | |
| Wu et al. | Deadline-constrained cost optimization approaches for workflow scheduling in clouds | |
| Zeng et al. | Quality driven web services composition | |
| Wagner et al. | QoS-aware automatic service composition by applying functional clustering | |
| Chen et al. | Multi-objective service composition with QoS dependencies | |
| CN104850727B (en) | Distributed big data system risk appraisal procedure based on Cloud focus theory | |
| Mostafa et al. | Fog resource selection using historical executions | |
| CN109408046B (en) | Shortest-path Web service combination method based on graph | |
| Chattopadhyay et al. | A fast and scalable mechanism for web service composition | |
| Wang et al. | Integrating trust with user preference for effective web service composition | |
| Niu et al. | Towards the optimality of QoS-aware web service composition with uncertainty | |
| da Silva et al. | Evolutionary computation for automatic web service composition: an indirect representation approach | |
| Qin et al. | AILS: A budget-constrained adaptive iterated local search for workflow scheduling in cloud environment | |
| Kil et al. | Transaction history-based web service composition for uncertain QoS | |
| Faruk et al. | A genetic PSO algorithm with QoS-aware cluster cloud service composition | |
| Yuan et al. | Resource-aware federated neural architecture search over heterogeneous mobile devices | |
| Shahrokh et al. | QoS-based web service composition applying an improved genetic algorithm (IGA) method | |
| Xu et al. | Efficient composition of semantic web services with end-to-end QoS optimization | |
| Liao et al. | A multi-objective service selection algorithm for service composition | |
| Ardagna et al. | Dynamic web service composition with QoS constraints | |
| Zhao et al. | Distributed resource allocation for synchronous fork and join processing networks | |
| Horng et al. | Merging artificial immune system and ordinal optimization for solving the optimal buffer resource allocation of production line | |
| Xie et al. | QoS driven web services evolution | |
| Shin et al. | An automated composition of information web services based on functional semantics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |