CN111431731B - Apparatus and method for acquiring system configuration of distributed system - Google Patents
Apparatus and method for acquiring system configuration of distributed system Download PDFInfo
- Publication number
- CN111431731B CN111431731B CN202010018158.7A CN202010018158A CN111431731B CN 111431731 B CN111431731 B CN 111431731B CN 202010018158 A CN202010018158 A CN 202010018158A CN 111431731 B CN111431731 B CN 111431731B
- Authority
- CN
- China
- Prior art keywords
- communication
- processing
- allocation
- functions
- function
- 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
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000004891 communication Methods 0.000 claims abstract description 185
- 230000006870 function Effects 0.000 claims abstract description 130
- 230000005540 biological transmission Effects 0.000 claims description 6
- 230000002068 genetic effect Effects 0.000 claims description 3
- 230000003190 augmentative effect Effects 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 claims 3
- 238000004590 computer program Methods 0.000 claims 1
- 238000004519 manufacturing process Methods 0.000 claims 1
- 238000005457 optimization Methods 0.000 description 13
- 101000667310 Macrovipera lebetina Venom serine proteinase-like protein 2 Proteins 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 101100510289 Caenorhabditis elegans klp-3 gene Proteins 0.000 description 1
- 101100181069 Drosophila melanogaster Klp61F gene Proteins 0.000 description 1
- 101000971697 Homo sapiens Kinesin-like protein KIF1B Proteins 0.000 description 1
- 101150086416 KIF15 gene Proteins 0.000 description 1
- 102100027630 Kinesin-like protein KIF15 Human genes 0.000 description 1
- 101100510288 Schizosaccharomyces pombe (strain 972 / ATCC 24843) klp2 gene Proteins 0.000 description 1
- 101100288056 Xenopus laevis kif15-a gene Proteins 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000000712 assembly Effects 0.000 description 1
- 238000000429 assembly Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/28—Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
- H04L12/2803—Home automation networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/085—Retrieval of network configuration; Tracking network configuration history
- H04L41/0853—Retrieval of network configuration; Tracking network configuration history by actively collecting configuration information or by backing up configuration information
Abstract
The invention relates to a method for obtaining one or more optimized system configurations of a distributed system described by a technical model, wherein the technical model is defined by a processing node (21) and a communication connection (22) between the processing nodes (21), wherein the system configuration describes an allocation relation of functions (11) of a functional model (1) to the individual processing nodes (21) of the technical model (2) and an allocation relation of communication paths (12) to the communication connection (22) between the processing nodes (21), wherein the communication paths (12) for the allocation of functions are described by the requirements of a function for communication with a function allocated in the other processing nodes (21).
Description
Technical Field
The invention relates to a method for distributing the functionality of a functional model in a distributed system with a plurality of networked processing nodes. The invention relates in particular to measures for allocating functions to individual processing nodes in terms of optimal resource utilization.
Background
In designing a distributed system, a problem of assigning functions of a function model to respective processing nodes is known. Such allocation is typically performed manually. In this case, a configuration is produced which does not take full advantage of the possibilities of the distributed system and furthermore does not allow a degree of freedom in designing the processing nodes of the distributed system and their communication connections.
For example, in motor vehicles, a plurality of controllers are provided, which are connected to one another by means of suitable communication connections. The controllers include, for example, motor controllers, transmission controllers, driver assistance systems, and different control assemblies for distributed vehicle functions. The controller and the control assembly represent processing nodes in which the functions of the functional model are implemented as software, so that the controller can be assigned and perform the functions implemented in software.
Typically, in such distributed systems, functions are assigned to processing nodes that are in the vicinity of or assigned to an organization to be controlled by the relevant function. In addition to this, functions can be intuitively assigned to the respective processing nodes by an expert. This may lead to the following situation, namely: the available hardware capacity cannot be optimally (too much or too little) fully utilized and the data communication between the processing nodes must be provided with a higher bandwidth than is required in the other functional allocations if necessary for the communication connection. The more complex or extensive a distributed system, the more difficult it is to intuitively assign the functionality of the functional model to the individual processing nodes of the distributed system.
In the solutions known from the prior art, the optimization is limited to local observations that do not take into account all the possibilities of resource sharing. The solution of taking into account the resources of the communication connection in addition to the resources of the processing nodes is focused on finding the shortest communication path, which does not in any case lead to a suboptimal system utilization when viewed globally.
Disclosure of Invention
According to the invention, a method according to the invention for compiling an optimized system configuration of a distributed system described by a technical model and an apparatus according to the invention are provided.
Other designs are described in other sections of this disclosure.
According to a first aspect, a method is provided for obtaining one or more optimized system configurations of a distributed system described by a technical model, wherein the technical model is defined by communication connections between processing nodes, wherein the system configuration describes an allocation relation of functions of a functional model to the respective processing nodes of the technical model and an allocation relation of communication paths to communication connections between processing nodes, wherein the communication paths for the allocation of functions are described by communication requirements of a function to a function allocated in a further processing node, wherein the following steps are performed:
by means of
The method comprises the steps of firstly, allocating functions of the functional model to processing nodes of the technical model in function allocation;
-determining a communication path according to said function allocation;
allocating the communication paths to the communication connections of the technical model in a communication allocation, respectively, for obtaining a system configuration;
to obtain a plurality of system configurations,
-obtaining, for each of the obtained system configurations, a processing cost of the allocation of functions of all processing nodes and a communication cost of the allocation of communication between all processing nodes; and is also provided with
-determining one or more optimized system configurations by a selection from the acquired system configurations according to the processing costs and the communication costs.
The idea of the above-described method is that the functions of the functional model are assigned to the processing nodes of the distributed system described by the technical model by means of a suitable optimization method for obtaining an optimized system configuration that utilizes the available resources of both the processing nodes and the communication connections between the processing nodes according to the optimization objective. The optimization is performed in two steps by means of allowing allocation of functions of the functional model to processing nodes and by means of allowing allocation of communication paths between functions of the functional model to communication connections between processing nodes, wherein pareto-optimal solutions are obtained in an iterative manner.
In this way, a large number of possible system configurations can be evaluated in terms of optimal system utilization for obtaining an advantageous solution for the system configuration. In addition, the cost of the communication connection can also be taken into account with the overall costs based on the optimization method. All possible schemes for transmitting data or information between two processing nodes are evaluated.
In particular, with the above-described solution, it is also possible to find an optimized system configuration taking into account longer communication paths that meet the requirements and reduce the costs of the overall system design.
Furthermore, processing requirement parameters can be assigned to the functions of the functional model, wherein in the function assignment, the functions of the functional model are assigned to the processing nodes of the technical model in each case in such a way that the processing requirement parameters of the functions assigned to the respective processing nodes do not accordingly violate the boundary conditions predefined by the respective processing efficiency parameters.
In particular, the processing requirement parameters of each function can be predefined for one or more resource classes, wherein the resource classes, in particular for the processing node, may comprise one or more of the following classes: computing power, storage capacity, direct access to sensor data, classification according to functional security level, interface to actuators, and energy requirements.
Furthermore, communication request parameters can be assigned to the communication connections of the functional model, which are assigned to the communication paths of the communication connections comprising the technical model in the communication assignment, so that the communication request parameters of all the communication connections assigned to the processing nodes do not accordingly violate the boundary conditions predefined by the corresponding communication efficiency parameters.
In particular, for one or more resource classes, a communication requirement parameter can be predefined for each communication path between two functions in different processing nodes, wherein the resource classes in particular for the communication path comprise one or more of the following classes: bandwidth, minimum latency, type of medium (optical/electrical/wireless), topology, protocol (synchronous/asynchronous) and direction of transmission.
It can be provided that the determination of one or more system configurations is carried out as a function of the processing costs and the communication costs by: the total cost is determined from the processing cost and the communication cost and one or more system configurations are determined from the total cost.
Alternatively, the determination of one or more optimized system configurations can be implemented in terms of processing costs and communication costs by: a pareto set of system configurations is selected from the plurality of acquired system configurations.
In particular, the processing costs in one of the processing nodes can be derived from the processing cost function and from the implementation or utilization of one or more of the resource classes for the processing nodes, and/or the communication costs for one of the communication connections can be derived from the communication cost function and from the implementation or utilization of one or more of the resource classes for the communication connections. If the resource class is capacity based, the resource class has utilization, whereas for discrete resource classes, such as classification according to functional security, the resource class has degrees of implementation.
According to one embodiment, the utilization of one or more of the resource classes for the processing node can be determined accordingly by a difference between a sum of the processing requirement parameters of the functions allocated for the relevant processing node for the relevant resource class and the predefined processing efficiency parameters of the relevant processing node, and/or the utilization of one or more of the resource classes for the communication connection can be determined accordingly by a difference between a sum of the communication requirement parameters of the communication path allocated for the relevant communication connection for the relevant resource class and the predefined communication efficiency parameters of the relevant communication connection.
Furthermore, at least one of the processing cost functions and/or at least one of the communication cost functions can correspond to a bathtub-shaped cost function that delineates costs by utilization, assigning high costs to low and high utilization and low costs to medium utilization. The use of bathtub-shaped cost functions advantageously enables optimized allocation scenarios in which the use of processing nodes avoids disadvantageous hardware configurations in which the processing nodes experience too low a use and thus are not effectively utilized or experience too high a use and thus lose scalability and flexibility.
In addition, one or more additional processing nodes and one or more communication connections can be augmented to a technical model for developing one of the system configurations. After the iterative method steps have ended, unused processing nodes and/or communication nodes can be removed from the technical model.
According to one embodiment, the assignment of functions of the functional model to the functional assignment and/or the assignment of communication links to the communication assignment can be optimized, in particular, by the use of genetic algorithms.
According to another aspect, a method for creating a distributed system is provided, wherein the distributed system is created with a system configuration, which is determined as an optimized system configuration according to the above method.
According to another aspect, an apparatus is provided for acquiring one or more optimized system configurations of a distributed system described by a technical model, wherein the technical model is defined by communication connections between processing nodes, wherein the system configuration describes an allocation of functions of a functional model to the individual processing nodes of the technical model and an allocation of communication paths to communication connections between processing nodes, wherein the communication paths for the allocation of functions are described by communication requirements of a function to a function allocated in other processing nodes, wherein the apparatus is configured for: by means of
The method comprises the steps of firstly, allocating functions of the functional model to processing nodes of the technical model in function allocation;
-determining a communication path according to said function allocation;
allocating the communication paths to the communication connections of the technical model in a communication allocation, respectively, for obtaining a corresponding system configuration;
to obtain a plurality of system configurations,
-obtaining, for each of the obtained system configurations, a processing cost of the allocation of functions of all processing nodes and a communication cost of the allocation of communication between all processing nodes; and is also provided with
-determining the one or more optimized system configurations by a selection from the acquired system configurations according to the processing costs and the communication costs.
Drawings
The embodiments are explained in detail below with the aid of the figures. Wherein:
FIG. 1 illustrates a schematic diagram of a functional and distributed system processing node allocation problem;
FIG. 2 shows a flow chart illustrating a method for finding an optimized system configuration for a distributed system with a given set of functions; and is also provided with
Fig. 3 shows a schematic diagram of a bathtub-type cost function.
Detailed Description
A method for finding a system configuration is described below, wherein the functions of a functional model are assigned to the individual processing nodes of a distributed system described by a technical model, and the communication paths are assigned by communication connections between the processing nodes.
For this purpose, fig. 1 schematically shows the underlying problem. It comprises a functional model 1 with a set of interrelated functions 11. These functions 11 are at least partially designed such that they exchange data with one another via a communication relationship 12.
Furthermore, a technical model 2 is provided, which describes a distributed system with individual processing nodes 21, which can be in connection with each other via a communication connection 22. The processing node 21 can be a programmable controller or assembly to which no function 11 of the functional model 1 or one, more or all of its functions, respectively, can be assigned or which can perform these functions. A communication connection 22 can be provided between at least a part of the processing nodes 21.
Processing efficiency parameters VLP1, VLP2, VLP 3 … or communication efficiency parameters KLP, KLP2, KLP 3 … are assigned to the processing nodes 21 and to the communication connection 22 between the two processing nodes 21, respectively. The processing efficiency parameters VLP1, VLP2, VLP 3 … for the processing node 21 can relate to at least one of the following resource categories: computing power, storage capacity, direct access to sensor data, relative to actuationInterfaces to the device, classification according to the level of security on the function, other discrete parameters and energy requirements. Communication efficiency parameter KLP for communication connection 22 1 、KLP 2 、KLP 3 … can relate to at least one of the following resource categories: bandwidth, minimum latency, type of medium (optical/electrical/wireless), topology, protocol (synchronous/asynchronous) and direction of transmission.
The system configuration 3, for which one of the processing nodes 21 is assigned to each of the functions 11 of the functional model 1, is generated by means of an optimization method described below. The functions, when implemented, can then be implemented and executed in the respective processing node 21. Furthermore, communication paths between functions 11 implemented in different processing nodes 21 are to be defined.
The functions 11 of the functional model 1 are each used with processing requirement parameters VAP for the processing nodes 21 for the respective resource classes (indexes) 1 、VAP 2 、VAP 3 …, which specify the required computing power (demand for computing power), the required storage capacity (storage demand), the possible access to sensor data, the possible direct operability of the actuator and, if necessary, further processing requirement parameters for its execution. The processing requirement parameter VAP is a function of interest and specifies for each observed resource class the working capacity of the processing node 21 to be allocatedIs required for the respective requirements of (a).
Furthermore, the communication requirement parameter KAP 1 、KAP 2 、KAP 3 … defines requirements for communication between two (distributed) functions 11 implemented in different processing nodes 21, which relate to data reception and data transmission from and to other processing nodes 21 for the respective resource classes, like e.g. bandwidth, maximum allowed delay, necessary transmission direction and, if necessary, other communication requirement parameters. Thus, the communication requirement parameter KAP for related functions and related resource categories 1 、KAP 2 、KAP 3 … each illustrate a corresponding communication requirement for the communication connection 22.
The method is explained in detail below with the aid of the flow chart of fig. 2. The method is provided with information about the configuration of the technical model 2, which depicts the output structure of the processing nodes 21 in the distributed system, and about the functional model 1 with a set of functions 11 to be implemented, in an appropriate manner.
In step S1, the technical model 2 can be expanded first by: processing nodes 21 are added that are identical to or differ from the existing processing nodes in one or more of the processing efficiency parameters. These additional processing nodes 21 can be connected with one or more of the processing nodes 21 of the original technical model 2 with additional communication connections 22 of different types. The extent of the expansion depends on the problem and can also be completely eliminated. For example, a number of processing nodes 21 can be added to the technical model 2, which corresponds to approximately 30% to 100%, in particular 50%, of the number of processing nodes 21 of the original technical model 2. An extended technical model is thus obtained, which enables variable functional allocation.
In step S2, an allocation method is performed in which the individual functions 11 of the functional model 1 are allocated to the processing nodes 21. The allocation is performed in such a way that the processing requirement parameters VAP for the processing in the processing node 21 under observation, which are to be allocated to the function 11 of the processing node 21 concerned, are passed through 1 、VAP 2 、VAP 3 … is correspondingly lower than the corresponding processing efficiency parameter VLP of the allocated processing node 21 1 、VLP 2 、VLP 3 … (constraints). That is, the functions to be allocated to the respective processing nodes 21 can be performed with resources available in the processing nodes 21.
Search space reduction (Suchrumreduzierung) is utilized in the allocation of functions by: only such processing nodes meeting the requirements of the functions to be allocated are considered. Is thatNot in this case by handling the requirement parameter VAP 1 、VAP 2 、VAP 3 … and processing efficiency parameter VLP 1 、VLP 2 、VLP 3 …. An example of this would be the case where the allocation of functions is declared invalid once the utilization of the processing node with respect to the relevant resource parameter exceeds 100%. But other arbitrarily chosen limits (90%, 120% …) are also contemplated. Another example of this can be the following, namely: functions for functional security with the ASIL C classification are allocated only to processing nodes with ASIL C or ASIL D, not to ASIL a and ASIL B classifications.
Processing efficiency parameters VLP capable of being at processing nodes 21 for respective functions 11 1 、VLP 2 、VLP 3 … or Process requirement parameter VAP 1 、VAP 2 、VAP 3 Processing cost functions VK, VK for each processing efficiency parameter or processing requirement parameter in … 1 、VK 2 、VK 3 The allocation relation of functions to the processing nodes 21 of the extended technical model 2 is determined on the basis of …. The processing cost GVK is acquired here.
For the assigned number z of processing nodes, the number m of functions to be assigned, applies for the processing cost GVK:
wherein at said processing efficiency parameter VLP 1 、VLP 2 、VLP 3 … and for each processing node 21 for each capacity-based resource parameter, the following applies:
for each discrete resource parameter, e.g. indicating a need for the performance of the processing node 21, the following conditions are only applicable:
VLP i ＞＝VAP i 。
for example, one possible processing cost function VKi for one or more of the resource categories is shown in fig. 3. In this case, if the function 11 to be allocated to the processing node 21 for the relevant resource class reaches a medium utilization N for the relevant resource, a minimum cost is incurred for implementing a plurality of functions in the processing node 21 for the resource class. The utilization N can be defined as the ratio of the sum of the resource demands on the associated resource class, defined by the respective processing requirement parameter VAP of the function to be allocated, to the maximum resource availability for the associated resource class, indicated by the respective processing efficiency parameter. If the resource class is capacity based, the resource class has a utilization, whereas for discrete resource classes, like for example a classification according to functional security, the resource class has a degree of implementation. For example, the functions can each define a storage requirement as a processing requirement parameter, wherein the total storage requirement of the function, which is determined by the accumulation of the individual storage requirements, is evaluated with respect to the maximum available memory, which is specified by the processing efficiency parameter of the associated processing node.
A preferred processing cost function VK for one or more resource classes can be determined from a bathtub-shaped function that assigns a cost K to the utilization N. Here, high costs for low and high utilization N and low costs for resource utilization in medium utilization N are described. For these medium utilizations N, it can be assumed, for example, that the cost is particularly low between 30% and 80%, preferably between 40% and 60%.
By using the bathtub-shaped cost function, the utilization N of the processing node 21 for the resource class can be adjusted by the function 11 assigned to the associated processing node 21 to a range which enables good resource utilization and retention of the capacity of the processing node 21.
The allocation method can be performed by genetic algorithms or other heuristic methods with the aim of minimizing the processing costs GVK.
Next, for the functional allocation of the function 11 and the processing node 21 found, a communication path between the two functions, which is formed by one or more communication connections 22, is determined. In order to simplify the subsequent optimization, a search space reduction can be performed in step S3 for finding possible communication paths 12 according to the allocation of functions and processing nodes, by means of which communication can be utilized according to the communication requirements between two distributed functions.
This significantly reduces the degree of freedom of the existing communication connections 22 for allocating the required communication paths to the technical model 2, so that the optimization speed can be significantly increased and the convergence can be improved.
In a subsequent step S4, the requirements for the respective communication paths between the processing nodes 21 are respectively assigned to the communication connections 22. The communication path between processing nodes 21 is created by the communication requirements of each of the functions with another function implemented in the other processing nodes 21. The allocation is verified under given boundary conditions.
The allocation is performed in such a way that the communication requirement parameter KAP for the communication via the communication connection 22 is set by means of a communication path extending through the associated communication connection 22 1 、KAP 2 、KAP 3 …, all requirements produced accordingly meet the given boundary conditions, for example by: all required by the corresponding communication efficiency parameter KLP below the observed communication connection 22 1 、KLP 2 、KLP 3 …. Correspondingly, the communication between the two distributed functions 11 can be performed only via a communication path consisting of one or more communication connections 22, if its communication efficiency parameter KLP fulfils the following conditions:
where k corresponds to the set of communication connections 22 and j corresponds to the index of the resource class of the communication efficiency parameter KLP.
Communication path 12 andthe allocation relation of the communication connection 22 can be defined for the communication connection 22 or the communication requirement parameter KAP 1 、KAP 2 、KAP 3 .. efficiency parameter KLP 1 、KLP 2 、KLP 3 Communication cost function KK for each communication efficiency parameter in … 1 、KK 2 、KK 3 … on the basis of the above. Where the communication cost GKK is determined.
The method is applicable to:
the communication cost function KK for one or more resource categories can be determined from a bathtub-shaped function that assigns a cost K to the utilization N of the communication connection 22. The high cost for resource utilization with small and high utilization N and the low cost for resource utilization in medium utilization N are described herein. For these medium utilizations, it can be assumed, for example, that the cost is particularly low between 30% and 80%, preferably between 40% and 60%.
By using a bathtub shaped cost function, the utilization of the communication connection 22 of the resource class can be adjusted to a range that enables efficient resource utilization and reservation of the capacity of the communication connection 22 for future expansion.
The allocation method for the communication connection 22 can be performed by an evolutionary algorithm or other heuristic method.
The respective cost shares relating to the processing resources and the communication resources used are now obtained in an iterative manner for the different system configurations.
The total cost GK is acquired in step S5. The total cost GK is derived as a weighted sum of the processing cost GVK and the communication cost GKK: gk=gvk+k GKK, where k is a predetermined weighting factor.
The system configuration thus obtained is saved, if its assigned total cost is lower than the total cost of the previously stored system configuration, otherwise the previously stored system configuration is saved as the current system configuration.
Alternatively, the system configuration can be collected in a set of pareto optimal solutions in the sense of multi-objective optimization.
The optimization steps S2 to S5 can be repeatedly performed. The system configuration, i.e. the allocation of the functions 11 to the processing nodes 21, is changed here and the communication path via the communication connection 22 is defined accordingly.
For this purpose, an interrupt condition is checked in step S6. Possible interrupt conditions for system optimization can include, for example, at least one of the following interrupt conditions:
-reaching a calculated duration;
the reduced variation of the total cost GK of the one or more optimization objectives is below a predetermined threshold, in particular for a predetermined number of successive iterations;
-a certain number of iterations is reached; and is also provided with
A certain number of iterations during which the total cost is not further reduced;
a certain number of iterations during which no further pareto-optimal solution is found.
If the interrupt condition is met (alternative: yes), the method continues with step S7, otherwise the method jumps back to step S2.
The acquired system configuration with the lowest total cost GK or the acquired system configuration of the pareto set is adopted as the optimized system configuration.
Due to the expansion of the technical model it is possible that the processing node 21 remains unused, i.e. has no functions assigned to it. Furthermore, the communication connections can remain unused, i.e. no communication paths are allocated for these communication connections. Thus, in step S7, the unused processing nodes 21 and the unused communication connections 22 are now searched for and, if necessary, removed from the obtained system configuration.
The above method enables an overall optimization of the system configuration taking into account the processing resources in the processing nodes and the communication resources of the communication connection between the processing nodes. Furthermore, an optimal solution can be found within the complex problem of a large number of functions with functional models and a large number of also heterogeneous processing nodes, within which the complexity of the person can be detected.
Claims (15)
1. A method for obtaining one or more optimized system configurations of a distributed system described by a technical model, wherein the technical model is defined by a processing node (21) and a communication connection (22) between processing nodes (21), wherein the system configuration describes an allocation relation of functions (11) of a functional model (1) to the respective processing nodes (21) of the technical model (2) and an allocation relation of communication paths (12) to the communication connection (22) between processing nodes (21), wherein the communication paths (12) for the allocation of functions are described by communication requirements of one function (11) to one function allocated in the other processing node (21), wherein the following steps are implemented in an iterative manner:
by means of
o assigning (S3) the functions (11) of the functional model (1) to the processing nodes (21) of the technical model (2) in a function assignment;
o determining a communication path (12) from the function allocation;
-allocating (S4) the communication paths (12) in a communication allocation to communication connections (22) of the technical model (2) for obtaining a corresponding system configuration;
to obtain a plurality of system configurations,
-obtaining, for each of the obtained system configurations, a processing cost (GVK) of the functional allocation of all processing nodes (21) and a communication cost (GKK) of the communication allocation (GKK) between all processing nodes (21); and is also provided with
-determining (S5) one or more optimized system configurations by selection from the acquired system configurations according to the processing costs (GVK) and the communication costs (GKK).
2. The method according to claim 1, wherein the functions of the functional model (1) are each assigned a processing requirement parameter (VAP), wherein in the function assignment the functions (11) of the functional model (1) are each assigned to a processing node (21) of the technical model (2) in such a way that the processing requirement parameters (VAP) assigned to the functions (11) of the respective processing node (21) respectively do not violate the boundary conditions predefined by the respective processing efficiency parameter (VLP).
3. The method according to claim 2, wherein the processing requirement parameters (VAP) of each function (11) are predefined for one or more resource classes, wherein the resource classes for the processing node (21) comprise one or more of the following classes: computing power, storage capacity, direct access to sensor data, classification according to functional security level, interface to actuators, and energy requirements.
4. Method according to claim 1, wherein the communication connections (22) of the functional model (1) are each assigned a communication requirement parameter (KAP) which is assigned in the communication assignment to a communication path (12) formed by the communication connections (22) of the technical model (2) such that the communication requirement parameters (KAP) of all the communication connections (22) assigned to the processing node (21) respectively do not violate a boundary condition which is predefined by the corresponding communication efficiency parameter (KAP).
5. The method according to claim 4, wherein the communication requirement parameters are predefined for each communication path (12) between two functions (11) in different processing nodes (21) for one or more resource categories, wherein the resource categories for a communication path (12) comprise one or more of the following categories: bandwidth, minimum latency, type of medium, topology, protocol, and direction of transmission.
6. The method of any of claims 1 to 5, wherein the determination of one or more system configurations is implemented in accordance with a processing cost (GVK) and a communication cost (GKK), by: the total cost (GK) is determined from the processing cost (GVK) and the communication cost (GKK).
7. The method of any of claims 1 to 5, wherein the determination of the one or more optimized system configurations is implemented in accordance with a processing cost (GVK) and a communication cost (GKK), by: a pareto set of system configurations is selected from the plurality of acquired system configurations.
8. The method according to any one of claims 1 to 5, wherein the processing cost (GVK) is obtained in one of the processing nodes (21) as a function of a processing cost function (VK) and as a function of the utilization of one or more resource classes for the processing node (21), and/or wherein the communication cost (GKK) for one of the communication connections is obtained as a function of a communication cost function (KK) and as a function of the utilization of one or more resource classes for the communication connection (22).
9. The method according to claim 8, wherein the utilization of one or more of the resource classes for the processing node (21) is respectively generated by a difference between a sum of the processing requirement parameters of the functions (11) for the relevant resource class assigned to the relevant processing node (21) and a predefined processing efficiency parameter (VLP) of the relevant processing node (21), and/or
Wherein the utilization of one or more of the resource classes for the communication connection (22) is produced in each case by the difference between the sum of the communication requirement parameters (KAP) assigned to the associated communication connection (22) for the communication path (12) of the associated resource class and the predefined communication efficiency parameter (KAP) of the associated communication connection (22).
10. Method according to claim 8, wherein at least one of the processing cost functions (VK) and/or at least one of the communication cost functions (KK) corresponds to a bathtub-shaped cost function reflecting costs by utilization efficiency, assigning high costs to low and high utilization and low costs to medium utilization.
11. The method according to any one of claims 1 to 5, wherein one or more additional processing nodes (21) and one or more communication connections with respect to the additional processing nodes (21) are augmented to the technical model (2) for programming one of the system configurations, and wherein unused processing nodes (21) and/or communication connections (22) are removed from the technical model (2) after the end of the iterative method steps.
12. Method according to any of claims 1 to 5, wherein the allocation of functions (11) of the functional model (1) is optimized in a functional allocation and/or the allocation of communication connections (22) is optimized in a communication allocation by the use of genetic algorithms.
13. A method for producing a distributed system, wherein the distributed system is compiled according to a system configuration, which is determined as an optimized system configuration according to the method according to any of claims 1 to 12.
14. An apparatus for obtaining one or more optimized system configurations of a distributed system described by a technology model (2), wherein the technology model (2) is defined by a processing node (21) and a communication connection between processing nodes (21), wherein the system configuration describes an allocation relation of functions (11) of a functional model (1) to the respective processing nodes (21) of the technology model (2) and an allocation relation of communication paths (12) to communication connections (22) between the processing nodes (21), wherein the communication paths (12) for function allocation are described by a communication requirement of a function (11) to a function (11) allocated in the other processing nodes (21), wherein the apparatus is configured for:
by means of
o assigning the functions (11) of the functional model (1) to the processing nodes (21) of the technical model (2) in a function assignment;
o determining a communication path (12) from the function allocation;
-assigning the communication paths (12) to communication connections (22) of the technical model (2) in a communication assignment for obtaining a corresponding system configuration;
to obtain a plurality of system configurations,
-obtaining, for each of the obtained system configurations, a processing cost (GVK) of the functional allocation of all processing nodes (21) and a communication cost (GKK) of the communication allocation between all processing nodes (21); and is also provided with
-determining the one or more optimized system configurations by a selection from the acquired system configurations according to the processing costs (GVK) and the communication costs (GKK).
15. Electronic storage medium having stored thereon a computer program set up for carrying out all the steps of the method according to any one of claims 1 to 12.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DE102019200169.6A DE102019200169A1 (en) | 2019-01-09 | 2019-01-09 | Method and device for determining a system configuration of a distributed system |
| DE102019200169.6 | 2019-01-09 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111431731A CN111431731A (en) | 2020-07-17 |
| CN111431731B true CN111431731B (en) | 2023-12-22 |
Family
ID=71104527
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010018158.7A Active CN111431731B (en) | 2019-01-09 | 2020-01-08 | Apparatus and method for acquiring system configuration of distributed system |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111431731B (en) |
| DE (1) | DE102019200169A1 (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4769772A (en) * | 1985-02-28 | 1988-09-06 | Honeywell Bull, Inc. | Automated query optimization method using both global and parallel local optimizations for materialization access planning for distributed databases |
| CN101398871A (en) * | 1995-02-13 | 2009-04-01 | 英特特拉斯特技术公司 | Systems and methods for secure transaction management and electronic rights protection |
| CN107926078A (en) * | 2015-08-28 | 2018-04-17 | 高通股份有限公司 | Small cell edge calculations platform |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006344017A (en) * | 2005-06-09 | 2006-12-21 | Hitachi Ltd | Sensor network system and data processing method for sensor network system |
| US9934325B2 (en) * | 2014-10-20 | 2018-04-03 | Korean Institute Of Science And Technology Information | Method and apparatus for distributing graph data in distributed computing environment |
-
2019
- 2019-01-09 DE DE102019200169.6A patent/DE102019200169A1/en active Pending
-
2020
- 2020-01-08 CN CN202010018158.7A patent/CN111431731B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4769772A (en) * | 1985-02-28 | 1988-09-06 | Honeywell Bull, Inc. | Automated query optimization method using both global and parallel local optimizations for materialization access planning for distributed databases |
| CN101398871A (en) * | 1995-02-13 | 2009-04-01 | 英特特拉斯特技术公司 | Systems and methods for secure transaction management and electronic rights protection |
| CN107926078A (en) * | 2015-08-28 | 2018-04-17 | 高通股份有限公司 | Small cell edge calculations platform |
Also Published As
| Publication number | Publication date |
|---|---|
| DE102019200169A1 (en) | 2020-07-09 |
| CN111431731A (en) | 2020-07-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109451540B (en) | Resource allocation method and equipment for network slices | |
| CN110069341B (en) | Method for scheduling tasks with dependency relationship configured according to needs by combining functions in edge computing | |
| CN107851043B (en) | Dynamic allocation of resource groups in a peripheral component interconnect express network | |
| WO2012020474A1 (en) | Job management device and job management method | |
| JP5285083B2 (en) | Method and apparatus for discovering topology in parallel | |
| CN109819032B (en) | Cloud robot task allocation method considering base station selection and computing migration in combined manner | |
| KR101382606B1 (en) | Apparatus and method for task mapping of hybrid optical networks on chip and hybrid optical networks on chip system using the same | |
| KR20160121380A (en) | Distributed file system using torus network and method for configuring and operating of the distributed file system using torus network | |
| CN111611076B (en) | Fair distribution method for mobile edge computing shared resources under task deployment constraint | |
| CN107360031B (en) | Virtual network mapping method based on optimized overhead-to-revenue ratio | |
| CN108347377B (en) | Data forwarding method and device | |
| CN111431731B (en) | Apparatus and method for acquiring system configuration of distributed system | |
| CN114385370A (en) | Memory allocation method, system, device and medium | |
| CN113535381A (en) | Service function chain mapping method, device and equipment based on improved genetic algorithm | |
| CN114546493A (en) | Core sharing method and device, processing core, electronic device and medium | |
| CN114073055B (en) | network controller | |
| CN111027689B (en) | Configuration method, device and computing system | |
| Ghosh et al. | Scalable QoS-based resource allocation in hierarchical networked environment | |
| Xu et al. | A mathematical model and dynamic programming based scheme for service function chain placement in NFV | |
| Laclau et al. | Predictive Network Configuration with Hierarchical Spectral Clustering for Software Defined Vehicles | |
| He et al. | An intent-based framework for vehicular edge computing | |
| CN116155750B (en) | Deep learning job resource placement method, system, equipment and storage medium | |
| CN112765056B (en) | Method, system, equipment and medium for reserving storage cluster LUN | |
| WO2016165072A1 (en) | Method and apparatus for deploying wired return link for wireless access point | |
| CN115348324B (en) | Method and device for determining optimized scheduling strategy and electronic equipment |
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 |