The Design of QoS Guarantee Strategy Framework for Networked Control System

The networked control system brings many advantages to the control system through network resource sharing, and introduces new challenges to the control circle. The complexity of the network control system is introduced by control system in the network of the characteristics of decision, such as network induced delay, packet loss, single packet of data and multiple packet transmission, packet random sequence and network scheduling problem, this article called “web services digital constraint” [1]. Network control applications require the use of heterogeneous, diverse resources, in general use of these resources sharing, more requires collaboration between the various tasks, resource scheduling, task execution is complicated. The current network infrastructure and are based on “best effort” basis, quality of service is not fully guaranteed, multiple services collaboration use resources, performance and efficiency is also easy to cause the drop. Addressing these issues requires the introduction of the QoS mechanism. Ian Foster and other scholars have proposed the QoS framework based on the Globus Toolkit -- the Globus Architecture for Reservation and Allocation, GARA [2][3], which provides the middleware core layer and the end-to-end QoS guarantee for the service layer. GARA use “gramm for resource management, defines the generic description mechanism for all kinds of heterogeneous resources and provide unified control interface, but the lack of extensibility, implementation architecture can’t applicable to the general field of networked control request. Professor at the university of Cardiff Rashid QoS management framework - G - QoSM [4], the use of the resources of QoS information registration, discovery and selection of reserve resources, resource allocation and task scheduling adaptive adjustment, the framework is the complete lack of QoS semantic modeling ability, change can not be very good to the business environment. Zeng et al. [5][6] proposed the service composition middleware based on maximization of user satisfaction, and gave two strategies for local task-level service selection and global collaborative allocation. The local service chooses to use the metric and weighted two steps to measure the service in the multidimensional QoS attribute. Literature [7] points out that the heterogeneous network environment put forward requirements definition scheme based on a known as the elastic constraints, the multi-objective problem is modeled as a compound option of knapsack problem, based on dynamic programming algorithm for dynamic constraint, but it is difficult to clearly determine the application of customers’ needs. S.S hakkottai et al. [8] in Ad Hoc wireless self-organizing network research, puts forward the idea of “cross layer design” in comprehensive considers the relationship of the protocol stack layers and retain the original layered protocol stack structure, on the basis of breaking the TCP/IP network hierarchy middle layer and the communication between the limit, supplement and designed a new type of network protocol stack, allowing interaction between the layers of communication protocol stacks. Xian university of electronic science and technology of yi-lin chang and others [9] in the cross layer design, on the basis of the five layers of typical network architecture was proposed and “cross-layer perception” concept, will be with the idea of cross-layer design, promotion to the network on the DE facto standard five levels. 26/5000 Most current network protocol implementations provide only partial QoS functionality. Networked control systems is to run over in the basic network services, the simple network QoS mechanism can guarantee the QoS requirements, and the existing QoS guarantee strategy are based on “the best” service mechanism, and cannot provide components oriented middleware platform support.

This chapter first analysis of networked control environment context factors, using the method of side oriented QoS based semantic model of real-time system, reconstruction of networked control system service quality in the hierarchical structure and establish the corresponding management model, by extending the object management group interface definition language, will establish the QoS management model integrating into the kernel of the middleware services, and construct the field oriented control service QoS management framework, finally through the simulation experiments verify the QoS management mechanism of effectiveness.

The structure of this paper is as follows: section 2 introduces the QoS semantic model and QoS service model for network-oriented control; Section 3 introduces QoS scheduling framework design; Section 4 introduces the QoS component access algorithm and QoS component deployment algorithm. Section 5 carries out the simulation experiment to verify the above strategy framework and the effectiveness of the algorithm.

This section presents a QoS semantic model, its purpose is to establish an open, extensible to define, QoS data exchange and interoperation model, does not allow the presentation logic to QoS number as the QoS parameters, QoS management level, the change of QoS management mechanism and fundamental changes, decouple the presentation logic and processing logic, so as to better adapt to the various business requirements under the networked control systems. The semantic model is composed of QoS model and QoS management model.

A.

The metamodel

QoS metamodel defines QoS said system required by the atomic data structure, it said in the QoS system plays a basic structural role, not only to enhance the QoS rigor of said system, and lay a foundation for the expansion of the QoS definition. The QoS model representation system is shown in figure 1 below.

Figure 1.

Model Root Element is the Root Element of the inheritance hierarchy in the metamodel, which is the most basic abstract block.

Identifier is the string that the model element identifies itself, which is unique in the entire namespace.

category : abstract building blocks used to derive data types and class 2 objects. It is an aggregation of methods and properties.

DataType which defines the basic data types used to describe QoS, construct data types, template data types, and custom data types.

Class: an abstraction of the collection of objects of the same behavior, attributes, relationships, and semantics.

Schema: a framework consisting of one or more classes.

Qualifier is the specification and constraint of elements such as class, attribute and its outline.

Event can be located in time and space and can take place with practical significance.

Rule is a policy used to indicate a class’s status migration when an event occurs.

B.

QoS management model

The control application is a set of application artifacts with QoS characteristics. The component instance is the physical unit that is deployed to the instantiation in the component framework, and the runtime component instance can be represented as a process, thread, or other autonomous running unit in the component container environment. An application Configuration Graph, which is called a QoS component, is made up of abstract examples.

Figure 2.

From the component instance to the directed edge, where the output is input. The QoS feature of the component is shown as follows.

Q o S C o m p o n e n t = ( Q o S d e m a n d , R e s , Q o S p r o v i d e ) Q o S d e m a n d = [ q 1 d e m , q 2 d e m , ⋯ q m d e m ] , q i d e m ∈ d o u b l e , i = 1 , ⋯ , m R e s = [ q 1 r e s , q 2 r e s , ⋯ , q k r e s ] , q i r e s ∈ d o u b l e , i = 1 , ⋯ , k Q o S p r o v i d e = [ q 1 p r o , q 2 p r o , ⋯ , q n p r o ] , q i p r o ∈ d o u b l e , i = 1 , ⋯ , n

In the above equation, the values of QoS attribute values are all real types, in which the QoS attribute list is represented and required by the component instance respectively. The other is the list of system resources occupied by the component instance runtime, which is provided by the system resource management service.

In this section, the simulation experiment is used to verify the core algorithm of the QoS management framework and related hierarchical strategy design.

Definition 1: QoS Requirements meeting Rate (QRMR: QoS Requirements Meet Rate), assuming that in all service application requests, the service application is successful, the QoS demand satisfaction Rate is defined as: Q R M R = S u c c e s s r e q u e s t T o t a l r e q u e s t

The simulation experiment of this paper compares QRMR with different component selection algorithms, as shown in fig. 6. The simulation experiment environment consists of four PCS networked with QoS component container.

Figure 6.

QoS simulation experiment environment hypothesis, influencing QRMR factors for resource constraints and service resource requirements. Simulation experiment, repeatedly randomly from every seconds to create a composite service request, the simulated four composite service function configuration diagram of components and service components deployed as shown in figure 7, and after a 5-1 calculated feasible QOS configuration resource requirements as shown in table 1.

Figure 7.

TABLE 1.

In the experiment, the QoS component only reserved management for the CPU resources, respectively representing the CPU resources.

The QRMR of the statistical system is simulated every 1 second. The duration of each successful service is set in seconds, and when the service is completed, the resource will be recovered by the corresponding QoS component resource management yuan service. In the experiments Capability _r1, Capability _r2, Capability _r3 and Capability _r4 The size is set to 100%. In the component access algorithm, r ₁, r ₂, r ₃, r ₄ The weights of resources are set respectively Q _r1 = 0.5, Q _r2 = 0.4, Q _r3 = 0.35, Q _r4 = 0.45.

Therefore, simulation negotiation algorithm is obtained by using the exhaustive method of integer programming optimal solution. Figure 8 shows the QRMR changed over time. Can see from the result of the experiment at the beginning of the simulation of two kinds of algorithm can achieve higher success rate, but as the simulation process, the heuristic member access algorithm QRMR meet rate is around 0.3, more than the basic component selection algorithm. The results show that the heuristic adjust weight and through service resources through negotiation methods, QRMR can improve.

Figure 8.

Can see from the result of the experiment at the beginning of the simulation of two kinds of algorithm can achieve higher success rate, but as the simulation process, the heuristic member access algorithm QRMR meet rate is around 0.3, more than the basic component selection algorithm. The results show that QRMR can be improved by heuristic regulation of resource weight and through service resource negotiation.

First introduces the basic requirements for networked control systems of QoS, shows an open and scalable QoS semantic model, its constituent elements including contract for QoS parameters, QoS, QoS model, specification of QoS, QoS parameters, etc., makes the service components have the ability to adapted to network resources. Based on the analysis of QoS attribute of service component, the establishment process of component service is described, and a dynamic algorithm of component service group is proposed. At the same time, the design for the application layer, middle layer and resource layer and network layer QoS strategy, finally designs the dynamic service components based on resource distribution of container deployment simulation experiment, this section is verified the proposed QoS goal-driven dynamic deployment effectiveness of the algorithm.