Abstract
In order to realize rapid service-oriented construction of resources and efficient supply-demand matching in avionics systems, an ontology-based resource modeling and matching framework in avionics systems is proposed. First, this paper constructs a hierarchical resource organization framework based on virtual resource pool, covering dispersed and heterogeneous avionics resources and shielding the differences and diversity of resources; next, introduces ontology technology into resource modeling and encapsulates resources as services using the idea of “resource as a service,” facilitating the access and control of resources; in addition, proposes a matching framework which takes into account the energy consumption, in which it firstly tries to find a list of resources which can meet task requirements by a stepwise intelligent matching strategy based on semantic similarity, and the resources in the list will then be sorted according to the energy consumption. Finally, a radar detection case is taken to demonstrate that the proposed model and method are feasible and effective.
1. Introduction
With the rapid development of avionics systems, the resource management of avionics systems is particularly important. Resource modeling and matching is the core technology of resource management in avionics systems. Avionics systems are composed of a large number of heterogeneous resources, which are interconnected by high-speed networks [1, 2]. In avionics systems, resources may have different descriptive information using different formats. They may also be indexed using different keywords referring to various characteristics. This imposes several difficulties for the resource management and configuration of avionics systems [2]. With the systematic development of avionics systems, in order to shield the differentiation and diversification of the resources, make comprehensive use of various resources, and collaboratively improve the ability to perform tasks, efficient resource modeling and matching methods are the premise of rapid and accurate sharing of avionics resources [3].
In avionics systems, resource modeling is the formal description of resources [4], so that resources have uniform semantics, accurate orientation, and unified understanding [5, 6]. Uniform semantics refers to the ability of resources to be stored in a uniform manner; accurate orientation means that upper-level applications do not need to care about the physical distribution of resources; and unified understanding means the structural model used here to describe the capabilities of resources in a unified way [7]. Based on the resource modeling, resource matching methods have also received widespread attention. In avionics systems, resource matching is to match the personalized task requirements with a variety of resources to find the resources that can best meet the needs [8].
Furthermore, resource modeling and matching becomes more challenging when an energy-related issue is added in this problem [9, 10]. On the one hand, effective resource allocation must meet the quality of service (QoS) requirements of all tasks. In addition, the meeting of the tasks’ requirements must ensure the lowest energy consumption [11]. Therefore, reducing the energy consumption does not only save a significant amount of money and improves the system reliability but also helps protecting our environment [12].
At present, the resource modeling methods proposed by domestic and foreign scholars mainly summarized into three categories [7, 8], which are the traditional ER model, the object-oriented model, and the ontology-based model. Among them, ontology model is a popular method at present. The first two methods are only a design method and solution, while ontology focuses on the description of the objective world and is committed to forming a unified standard description of the objective world. Many domestic and foreign scholars have done a lot of research on resource modelling and matching. Daouadji et al. [7] proposed an ontology-based resource description framework and an extensible resource discovery method to realize the virtual management of resources and consider the availability of energy. This method emphasizes the relationship between resources and energy and lacks comprehensive modelling and retrieval of resource attributes. Hu [8] proposed an ontology-based modeling method for domain model and used a semantic matching algorithm to implement the management, sharing, and semantic retrieval of manufacturing resources. But the resource servitization and the energy consumption are not considered. Li et al. [13] proposed an owl-based service description model and a multilevel intelligent matching method in the cloud manufacturing environment to realize the matching of manufacturing cloud services. Hu [14] used a three-level description mechanism to describe cloud manufacturing resources and services and used fuzzy similarity algorithm to match cloud manufacturing requirements and services. Zhuang [15] introduced a metadata-based service-oriented modeling and encapsulation method for machine tool resources and solved the supply-demand matching of machine tool resources in cloud manufacturing environment through function matching and performance matching, respectively. The literature [13–15] also lacks consideration of energy consumption. In addition, some research has suggested handling semantic aspect [16, 17] based on Web Service Resource Framework (WSRF) specification; however, WSRF will unlikely be appropriate for managing heterogeneous environments.
On the basis of the above researches, it can be seen that the research on ontology-based modeling and matching has developed by leaps and bounds. However, many resource management systems still do not support semantic resource modeling and matching, or lack consideration of energy consumption. There is still no better way to solve the problems of fast service-oriented construction and efficient supply-demand matching of resources for avionics systems yet, and the energy-related aspects and semantics expressing relationships among resources in terms of energy consumption interdependence are not taken into account. We should combine the characteristics of green resources to research and develop suitable methods for resource modeling and matching in avionics systems. Therefore, this paper proposes an ontology-based resource modeling and matching framework in avionics systems. The contributions of this paper are summarized as follows. (1)This paper proposes a hierarchical and modular resource organizational framework based on virtual resource pool which contributes to shield the differentiation and diversification of resources. In order to facilitate the access and control of resources, the ontology technology is introduced into resource modelling, and service-oriented encapsulation is used to abstract the dispersed and heterogeneous resources into a service. It lays a foundation for the intelligent matching of avionics resource services(2)For the purpose of simplifying resource matching process and improving matching efficiency as well as the green energy usage, this paper provides a resource matching framework in terms of energy consumption which match the attribute parameters by semantic similarity algorithms, it firstly tries to find a list of resources which can meet task requirements, and the resources in the list will then be sorted according to the energy consumption; finally, the most appropriate resource is returned to the task requester
The rest of this paper is organized by the following structure. First, Section 2 constructs a three-tier resource organizational framework based on virtual resource pool. Next, Section 3 introduces the ontology technology into resource modeling. Furthermore, Section 4 is the main part of this paper, which introduces our matching framework and the correlative methods in detail. In addition, Section 5 evaluates and discusses the method in this paper through experimentations. Finally, Section 6 summarizes the experimental results of this paper and explores prospects for future work.
2. Resource Organizational Framework Based on Virtual Resource Pool
In order to achieve dynamic and efficient management and control of resources and high-speed, real-time, distributed processing and sharing of massive information, this paper constructs a three-tier resource organizational framework based on virtual resource pool; the resource organizational framework here has physical resource layer, virtual resource layer and application layer [18] (see Figure 1).
The resource organization process can be divided into two stages. The first one stage is to make the physical units decoupled, determine the resource granularity, carry out the resource description and service encapsulation, and construct the virtual resource pool. The other stage is optimizing the management, combination, and matching of different resources and providing a unified resource interface for the upper application layer. As a result, the application layer sends requests to the physical resource layer through the virtual resource layer and obtains the request results.
The management approach for resources of avionics systems requires a flexible and extensible resource modeling support. The interdependence between application layer and physical resource layer as well as resource sharing and combination capabilities must be represented semantically. One of the appropriate solutions for such a model is using a particular ontology [7].
3. Unified Modeling of Resources
In order to facilitate the access and control of resources, the ontology technology is first introduced into the resource modeling, and then, the concept of “resource as a service” is used to encapsulate the resources as a service, which lays a foundation for the intelligent matching of avionics resource services.
3.1. Ontology-Based Modeling
Ontologies (often also referred as domain model) are generally defined as a “representation of a shared conceptualization of particular domain.” It helps modeling a world phenomenon by strictly defining its relevant concepts and relationships between the concepts, thereby providing a common understanding of domain knowledge and achieving knowledge sharing and reuse [19]. Therefore, ontology-based resource modeling can better support resource sharing. Each resource in avionics systems has its own attributes and behaviours; some of these properties and behaviours allow resource to perform technical operations. All of the resources and their properties have certain ranges and constraints. They can be for example technical properties such as maximum detection distance or detection range of the radar or environmental constraints like maximum allowed humidity and temperature. Based on the analysis of the characteristics of resources in avionics systems, the resource information of avionics systems is described abstractly (see Figure 2), including basic attributes, function attributes, status attributes, access attributes, and capability attributes.
Moreover, there is a relationship between resource and energy that defines the kind of energy consumed by resources, e.g., energy-saving or nonenergy-saving. Nonenergy-saving is more costly than energy-saving. In the context of avionics systems, we define energy cost as energy consumption which estimates the energy consumed by the execution of the task [20, 21].
On this basis, the resource ontology in avionics systems can be formally described as follows.
: the basic attributes are the general description of the basic parameters of a resource, which mainly includes the resources id, name, category, company name, manufacturer contact way, and other basic information.
: function attributes represent function name, type, and main function description. Different resources can accomplish different functions.
: the status attributes describe the current status of resources in avionics systems, which mainly include several state types such as high load, normal load, low load, idle, and reserved.
: interface attributes define the access mode provided by resources in avionics systems, including port ID, port address, port type, communication protocol, and data format.
: the capability of a resource is determined by the status, type, spatio-temporal constraints, and tasks [22]. The capability of resources is closely related to the task and cannot exist independently of the task.
3.2. Service-Oriented Encapsulation
In the face of the large-scale and diverse task requirements of avionics systems, resources are characterized by distribution, heterogeneity, and dynamic. In traditional avionics systems, resources and tasks are tightly coupled and bound, and there are barriers to interconnection and interoperability [23].
Therefore, a service-oriented encapsulation method for avionics resources is proposed to abstract the dispersed and heterogeneous resources into a service, thus reflecting the core concept of “resource as service” of avionics systems [24]. Resource service-oriented encapsulation is not only a linear superposition of existing resources, but can form a virtual service that can complete tasks. The resource service ontology of avionics systems is shown in below (see Figure 3).
On this basis, the ontology-based resource service for avionics systems can be formalized as:
: the basic attributes are the general description of the basic parameters of a service, which mainly includes the service id, service name, service type, service address, and other basic information.
: data attributes are the unified encapsulation of resources, which include resource id, resource function descriptions, status, and capability information. As resource services are still based on resources, in addition to inheriting various attributes of resources, some new attributes will also be generated, such as service time, cost, security, reliability, and maintainability.
: the access attributes define the access mode provided by the service in avionics systems, including port id, port type, port address, communication protocol, and data format.
In this paper, the semantic description of resources is implemented using web ontology language (OWL) [25], which is considered the best representation for ontology. The OWL [19, 25] has become the widely used ontology language for its benefits of the well-defined syntax, efficient reasoning support, formal semantics and the full expression ability, and so on. It can provide more semantic descriptions than XML, RDF, and RDFS through the increase of some modeling primitives (e.g., owl: class, owl: datatype-property, and owl: object-property), greatly improving the expressive ability of the model.
4. Intelligent Matching Method Based on Semantic Similarity
Resource matching in the context of avionics systems is massive and complex. In order to simplify the matching process and improve the matching efficiency, this paper builds a matching framework between task requirements and services on the basis of resource ontology modeling and then proposes a stepwise intelligent matching strategy based on semantic similarity using semantic similarity algorithms to match their attribute parameters, at last return the lowest energy consumption resource to task requester.
4.1. Intelligent Matching Framework
In avionics systems, the matching process of tasks and services is the selection process that the task requester selects the service provided by the service provider according to its requirements. By semantically matching task requirements and resource services, the matching algorithm firstly tries to find a list of resources which can meet task requirements, the resources in the list will then be sorted according to the energy consumption; finally, the most appropriate resource is returned to the requester [26].
Among them, the task ontology model [27–29] is represented as below. where is the set of basic information properties for the task, including task id, task length, task type, task priority, and task description. is the set of service attributes requested by the task, which includes service time, QoS, security, reliability, and maintainability. is the set of constraint properties associated with the task, including start time, end time, as well as loads.
As shown in Figure 4, the intelligent matching method based on semantic similarity includes three steps: basic information matching, functional information matching, and quality information matching. The matching similarity value of each step is calculated according to the semantic similarity algorithms, and then, a list of resources which can meet task requirements could be obtained after the first three steps and finally calculate the energy consumption of the current resources, and the resource with the lowest energy consumption is fed back to the task requester as the matching result. The larger the matching similarity value, the higher the similarity, and vice versa.
4.2. Intelligent Matching Method Based on Semantic Similarity Algorithms
According to the diversity and complexity of resource services in avionics systems, the service name and service description usually appear in the form of phrases, sentences, or concepts when describing service attributes, so keyword similarity calculation is usually used when matching such subattributes [14]. For service time, service reliability, etc. are usually expressed in numerical form; thus, parameters similarity calculation is usually used for matching such subattributes [13]. However, subattributes such as service security and service credibility are usually described by high, very high, general, very low, and low, which cannot be expressed by precise numerical data. As a result, fuzzy similarity calculation is required [14], and fuzzy language is converted into corresponding fuzzy value through intuitionistic fuzzy function.
4.2.1. Keyword Similarity Calculation
First, set the two sentences of the task requester and the service provider as vectors, respectively.
Then, construct a similarity matrix as Equation (5), and represents the semantic similarity of keywords .
Finally, take the maximum value of each row and perform normalization computing to obtain the similarity value as
4.2.2. Parameter Similarity Calculation
Set the two parameters of task requester and the service provider to and , respectively, and then, the similarity between any two parameters can be calculated as
4.2.3. Fuzzy Similarity Calculation
First, the quality of services of the task requester or the service provider is converted into a value using the following table (see Table 1), and then, the fuzzy similarity is obtained through parameter similarity calculation.
Among them, represents the membership degree, represents the non-membership degree, and represents the uncertainty degree given by users. At last, the following Equation (8) is used to calculate the fuzzy value.
4.2.4. Energy Consumption Calculation
Each resource has the computing capabilities and a power consumption ; the energy consumption usually varies with time and power consumption when a task is running on a resource [30, 31]. The energy consumption of resource is computed as follows: where is the length of task , measured by million instructions (MI), and is computing capabilities of resource , measured by millions of instructions per second (MIPS).
According to the matching algorithm research in the above several aspects, it can be concluded that service matching in avionics systems is a gradually accurate process. The specific matching process is as follows:
Step 1. Basic information matching. Match the basic information of the task requirements and the basic information of the service provider by semantic similarity algorithms. If the similarity is greater than or equal to the basic information matching threshold, it means that this service satisfies the task’s requirements for basic information and then goes to Step 2. Otherwise, the matching of the service is stopped, and a matching failure message is returned to the task requester.
Step 2. Functional information matching. Match the functional information of the task requirements and the functional information of the service provider by semantic similarity algorithms. If the similarity is greater than or equal to the functional information matching threshold, it means that this service satisfies the task’s requirements for functional information, and then goes to Step 3. Otherwise, the matching of the service is stopped, and a matching failure message is returned to the task requester.
Step 3. Quality information matching. Match the quality information of the task requirements and the quality information of the service provider by semantic similarity algorithms. If the similarity is greater than or equal to the quality information matching threshold, it means that this service satisfies the task’s requirements for quality information, and then goes to Step 4. Otherwise, the matching of the service is stopped, and a matching failure message is returned to the task requester.
Step 4. Calculate the energy consumption of the current resources. A list of resources can be obtained if the matching conditions in the first three steps are all satisfied, then calculate the energy consumption of the current resources in the resources list.
Step 5. Return the matching results. The resource with the lowest energy consumption is fed back to the task requester as the matching result.
5. Experimental Results and Analysis
In order to test the effectiveness and correctness of the proposed method, the simulation verification platform of avionics systems is established by simulation technology [32]. The experimental platform is a computer equipped with Intel (R) Core (TM) i7-10700H CPU 2.90 GHz, 16 GB RAM, and 64 bit Windows 10 operating system. The simulation program written in C language is used to realize the calling of services in avionics systems.
5.1. Radar Ontology Modelling Case
In order to display the resource ontology model in the avionics systems, the ontology editor tool protégé 5.5.0 [33] is used to build the resource model. Take the warning radar [34] as an example to carry out the service-oriented encapsulation, and some examples of the resource ontology model are shown by means of graphic display (see Figure 5).
Figure 5 intuitively shows the ontology modeling process of the warning radar. Through the analysis of attribute information composition, service-oriented modeling, and the encapsulation test of the warning radar, the search service and the detection service are finally generated.
5.2. Intelligent Matching Method Test
In the experiment, we take the detection function of the radar as an example. The important information of the radar is extracted as the requirement information and tested in the prototype of the virtual resource pool containing 6 virtual radar services. Suppose T1 is the task requirement and S1-S6 are six services. See Table 2 for detailed description. Here, we set the task length of to 150 MI, set the capabilities of S1-S6 to {10.11, 60.72, 27.40, 82.78, 62.65, 53.19} MIPS, and set the power of S1-S6 to {120, 180, 135, 145, 162, 155} watts, and set the weights of basic information, functional information and service quality to be equal and set to 1/3.
In order to narrow down the selection of tasks to precisely match the appropriate service, we used matching thresholds in our simulations. When the selection of the matching threshold is too small, we may get a wide range of service candidate sets; when the selection of the threshold is too large, we may not even get a resource that meets the requirements. Therefore, we either set the threshold to a fixed constant using past experience or use the related dynamic method to calculate the weights of each attribute. In this paper, for the convenience of experiments, we use past empirical values to set the matching threshold, and we set the matching thresholds for each step to be 0.90, 0.80, and 0.80, respectively.
Next, the step-by-step matching solution is carried out as follows:
Step 1. Basic information matching. The basic information matching is the matching of service name and service description, and the keyword similarity is calculated according to Equation (5) and Equation (6). Calculate similarity by , where . Then, get . It can be concluded that , and all meet the matching conditions and then go to Step 2. While , so matches failed.
Step 2. Functional information matching. In this experiment, the function information matching is the matching of service state, and the parameter similarity is calculated according to Equation (7). The service state ranges from 1 to 5 level, representing high load, normal load, low load, idle, and reserved. The required service state is low load, so the service satisfies the matching conditions if the service state is idle or low load. Calculate similarity by , then get . It can be concluded that , and all meet the matching conditions and then go to Step 3. While , so matches failed.
Step 3. Quality information matching. The quality information matching is the matching of security, reliability, and credibility. The similarity of reliability is calculated according to Equation (7) for parameter similarity, and the similarity of security and reliability is calculated according to Equation (8) for fuzzy similarity. The uncertainty degree is a set of random numbers in the interval . Calculate similarity by , where . Then, get , . It can be concluded that all meet the matching condition and then go to Step 4, while , , so match failed.
Step 4. Calculate the energy consumption of the current resources. Here, we do not need directly set the coefficient of energy consumption. Instead, a resource list consisting of S5 and S6 can be obtained after the first three steps, then calculate the energy consumption of the current resources S5 and S6, respectively, by Equation (9). Then, get , , and , so it can be concluded that is greener than , and then, go to Step 5.
Step 5. Return the matching results. Take as the final matching result and fed back to the task requester.
It can be seen from the above matching test that the stepwise intelligent matching algorithm proposed in this paper meets the actual requirements of the avionics systems and can solve the matching problem between task requirements and resource services when taking energy consumption into account.
5.3. Performance Test
To evaluate the performance of the proposed algorithm, we tested the average query time and the number of services in the list that meet the requirements in the resource pool containing 1000 services. We compare the proposed algorithm with the traditional keyword-based matching algorithm [35] and the matching method based on semantic similarity algorithm [26]. In order to offset the influence of data randomness and better reflect the performance of the algorithm, the final value is taken as the average of all values obtained in 20 tests. The specific test results are shown in below (see Table 3).
For the task requirements, as the number of services in the resource pool increases from 100 to 1000, Figures 6 and 7 show the performance of the algorithm in terms of average query time and the number of services in the list that meet the requirements, respectively. Figure 6 shows a significant improvement in the average query time, this is due to the fact that the more query time should be considered when the more services that are tested on. However, it can be seen that average query time of our proposed algorithm is smaller than that of the two algorithms, that is because that resources and requirements can be matched flexibly by means of step-by-step intelligent matching process and threshold setting for each step, so as to quickly filter out massive resources that do not meet the task requirements and improve the efficiency of the matching process. Figure 7 shows the number of services in the list that meet the matching conditions when the number of services changes. For the same number of servers, the number of services in the list that meet the requirements of the traditional keyword matching algorithm is less than that of the two algorithms, since the attributes of services are not considered fully. And the number of services in the list that meet the requirements of our proposed algorithm is less than the matching method based on semantic similarity algorithm; compared with the single attribute information matching algorithm, our proposed algorithm can get more accurate matching results. In addition, our algorithm introduced ontology modeling and service-oriented encapsulation to realize the servitization and efficient sharing of resources. Furthermore, our proposed algorithm also uses the OWL to implement semantic description of resources to facilitate resource retrieval. Furthermore, our proposed algorithm also considers energy consumption on the basis of matching method.
6. Conclusions
Ontology technology has been widely studied and applied. In order to realize rapid service-oriented construction of resources and efficient supply-demand matching in avionics systems, this paper proposes an ontology-based resource modeling and matching framework in avionics systems. First, a hierarchical and modular organizational framework for resources based on virtual resource pool is constructed, which shields the differentiation and diversification of resources, and facilitates the management and control of resources. Second, the ontology modeling and service-based encapsulation of resources lay the foundation for the intelligent matching of resources. Third, a matching framework between task requirements and services on the basis of resource ontology modeling is proposed and then proposes a stepwise intelligent matching method based on semantic similarity algorithms to match their attribute parameters, at last return the lowest energy consumption resource to task requester. Finally, the feasibility and effectiveness of the model and method are verified by experiments. In the future, we will focus on in-depth research on the problem of combination optimization of resources.
Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.
Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper.
Acknowledgments
The work presented in this paper was supported in part by the National Natural Science Foundation of China no. 62106202, the National Natural Science Foundation of China no. 62102316, and the Aeronautical Science Foundation of China no. 2020Z023053004.