CN1307549C - Component Management Method Supporting Pervasive Computing - Google Patents

Abstract

The invention discloses a component management method supporting pervasive computing. 1) In addition to the encapsulation of the general component itself, the component container is used to encapsulate the component twice, and the runtime environment of the component is provided to solve the problem of inconsistent granularity of the component; 2) The meta-information of the component runtime is managed through the component control block , which contains component description information and control information, each component corresponds to a component control block, which is a concentrated reflection of component dynamics and the basis for the system to identify and control components; 3) A component register is introduced to record All components that can be perceived by a certain device realize dynamic registration, replacement, addition and deletion of components. The invention has the advantage of further encapsulating on the basis of components, thereby supporting the characteristics of high heterogeneity, high mobility and high transparency under the pervasive computing environment. Make components of different granularities and different manufacturers well compatible; realize dynamic registration, replacement, addition and deletion of network components.

Description

Component Management Method Supporting Pervasive Computing

technical field

The invention relates to componentized software design, in particular to a component management method supporting pervasive computing.

technical background

Under the traditional development technology, large-scale software often has problems such as complex system analysis, long development cycle, high maintenance cost and difficulty in function expansion. Not only that, with the rapid development of network technology, software needs to have an open system structure and the ability to span multiple heterogeneous platforms. If the software is heavily dependent on the operating system and specific network services, the openness of the application system will be poor. In order to adapt to different application environments, developers have to develop different codes, which will cause a lot of duplication of work, which seriously affects the improvement of software production efficiency and software quality.

The emergence of component technology has solved various problems that have plagued the software engineering field for a long time. The so-called software component refers to a reusable software unit that can be used to construct other software. It can be encapsulated object classes, some functional modules, software frameworks, software system models, software documents, and so on. Software components do not depend on a certain system, can be replaced by functional components, and have practical functional significance. Component-based software development is to select the applicable components from the component library, and construct the application software through the assembly and control of the components. Component technology enables software to have good platform compatibility, structure openness, scale variability, system configurability and code reuse capabilities.

Broadly speaking, components have the following basic characteristics. 1. A component is an independently configurable unit, so the component must be self-contained; 2. The component emphasizes the separation from the environment and other components, so the implementation of the component is strictly encapsulated, and the outside world has no chance or need to know the implementation details inside the component; 3. Components can be compounded and used in an appropriate environment, so components need to provide clear interface specifications and can interact with the environment; 4. Components should not be persistent, that is, components have no individual-specific attributes, which means that components should not interact with the environment. Self copy difference.

From the above four characteristics, it can be seen that a component inherits the encapsulation characteristics of an object, but at the same time it is not limited to one object. It can encapsulate one or more classes, prototype objects or even processes inside, and its structure is flexible. Components highlight the characteristics of self-containment and containment,

Componentization is very important to the engineering of software development. Only with components can the engineering development of software be realized through the assembly and interconnection of components. Componentization is a cutting-edge software design idea, which promotes the development of the entire software industry. If we look back at the classic industrialization revolution, it is not difficult to draw some useful revelations: products with complex functions are composed of a large number of standard parts, and the parts are assembled into a finished product on the production line, and all parts play a role in the finished product. The more detailed the division of labor and the higher the degree of specialized production, the higher the overall production efficiency. The role of component technology is to apply the concepts of parts, production lines and assembly operations in the software industry. It can be predicted that software component technology is an inevitable development trend of the software industrialization revolution.

"Componentization" can be said to be a trend in the development of software technology across the century. It solves the shortcomings of traditional software that are difficult to reuse, maintain and update, and the system is not easy to customize. Although there are quite a few systems that support component-based software technology, there is no unified standard implementation plan. Each company designs components in different ways (such as COM, JAVA). They are all mainly to solve the independence of components, High reusability, portability, and maintainability of components.

With the development of computers, ubiquitous computing has gradually turned from an ideal to a reality. With the advancement of computer science, our living space will be filled with more and more wireless embedded devices, such as: handheld computers, BP machines, vehicle-mounted smart devices, notebook computers, watches, smart cards, smart phones, set-top boxes, POS vending machines and screen phones have different sizes, provide different functions, and exist in different forms. Some use a class as a component, some use a service as a component, and some even bundle several applications together as a component; such a situation results in heterogeneous components, making it difficult for components to be compatible. What is even more troublesome is that the components in the ubiquitous computing environment also have the characteristics of large number and strong mobility. A certain component may leave (or enter) the "line of sight" of a certain device at any time, which makes the component difficult to manage. These problems are yet to be resolved.

There are four requirements that a pervasive computing software system should meet:

1. Effective use of activity space: The meaning of activity space in ubiquitous computing is very broad; it can be a bedroom or a courtyard. All kinds of embedded devices must be organically integrated with the activity space. For example, in a room, embedded computing devices can control whether the heating and lights should be turned on according to the indoor temperature and brightness; similarly, changes in the environment can also cause changes in the functions of embedded devices.

2. Transparent to users: users do not need to care about the working process of embedded computing devices, but should be able to enjoy various high-quality services inadvertently. This requires embedded devices to be able to perceive the user's intentions and automatically adapt to various possible situations.

3. Local scalability: The computing activity space has a close interaction with the surrounding environment, but the available network bandwidth resources are very limited. Therefore, when an embedded computing device leaves, the active space must perceive these behaviors and cut off the connection with this device; similarly, when a computing device enters, the active space must get in touch with the new device. The purpose of this is to reduce the association between mobile users and devices that are far away, and allow mobile users to associate with devices within a small range as much as possible, thereby ensuring the effective utilization of computing resources.

4. Shielding of unbalanced loads: different environments have different numbers of computing devices. In conference rooms, offices, and laboratories, there may be a lot of embedded devices; while in some places with relatively closed information and few equipment, such as lawns, gardens, and football fields, there are only a handful of computing devices. Ubiquitous computing requires shielding these unbalanced situations and providing users with a unified environment.

From the perspective of software systems, in the ubiquitous computing environment, various small computing devices can achieve online transaction processing and enterprise computing under unstable communication conditions during the process of continuous movement or in a wide range of geographical distribution. Core data access. These small computing devices have a variety of communication means, such as mobile communication networks, infrared, Bluetooth, and satellites, and can be connected to the Internet or intranet, but this connection is not a fixed connection, but an intermittent connection. Therefore, ubiquitous computing requires new software architectures that allow users to use these devices for complex online transaction processing and information access.

Contents of the invention

The purpose of the present invention is to provide a component management method supporting ubiquitous computing.

1) In addition to the encapsulation of the general component itself, the component container is used to re-encapsulate the component and provide the runtime environment of the component to solve the problem of inconsistent component granularity;

2) Manage component runtime information through the component control block, which includes component description information and control information. Each component is uniquely corresponding to a component control block, which is a concentrated reflection of component dynamics. It is the identification and control of components by the system. basis for control;

3) A component register is introduced to record all components that a certain device can perceive, and realize dynamic registration, replacement, addition and deletion of components.

The invention has the advantage of further encapsulating on the basis of components, thereby supporting the characteristics of high heterogeneity, high mobility and high transparency under the pervasive computing environment. Make components of different granularity and different manufacturers well compatible; realize dynamic registration, replacement, addition and deletion of network components.

Description of drawings

Fig. 1 is a schematic diagram of component container encapsulation of components;

Fig. 2 is the structure of the component control block and a schematic illustration of an example;

Fig. 3 is a schematic diagram of component management by the component register;

Fig. 4 is a schematic diagram of perception space;

Fig. 5 is a schematic flowchart of the process of accessing components.

Detailed ways

The present invention is based on a brand-new software design idea - component-oriented design method, in our design "everything is a component". The invention mainly solves two problems: the heterogeneous problem of components and the mobility problem of components. The detailed description of the present invention is as follows:

Component management system software supporting pervasive computing should address the following key issues:

1. Discovery mechanism: When a user brings a new device into a space or adopts a new component on an existing device, how does this component and other available components and devices discover each other and how to communicate between them? What interactions are required. At the same time, the ubiquitous computing system is a gradually expanding system. This process should not require restarting the system or rewriting existing codes, which is also based on the discovery mechanism. There are already some technical foundations for this.

2. Self-adaptation: The resources owned by various devices in the ubiquitous computing system are different and changeable, from the smallest system to objects with embedded computing capabilities, to portable computing devices, and to components in the infrastructure. There are great differences in computing power, storage capacity, battery capacity, and interaction means; when devices move in different environments, the available wireless network bandwidth is also different. This requires solving the problem of self-adaptation. It includes two parts: transformation of content and transformation of interactive interface. The former solves the processing of data due to resource limitations in the process of data transmission; the latter is due to the different interaction mechanisms of devices, and needs to map the same interaction interface to different interaction mechanisms to complete.

3. Management of physical entities: Because in a pervasive computing system, objects in the physical world have meanings in the information space, the system software of pervasive computing must have certain management capabilities for these physical entities. The system software of ubiquitous computing is an operating system in physical space. The difference between it and the current operating system is that it needs to manage physical entities. The management here has two meanings: considering the physical entity as a resource, it is necessary to establish a representation of the location, structure, and function of these resources, and it needs to reflect the dynamic actual situation. This requires the support of context-aware computing. When physical entities are embedded with sensing, computing, and effecting capabilities, a high-level interface to them must be provided, which is equivalent to a traditional driver.

4. Coordination mechanism between components: Ubiquitous computing system is a distributed system, how to organize the complex interconnection, communication, and collaboration among various distributed components in a unified and organic way! This is the role of the coordination mechanism. The existing distributed component structures can be considered as a coordination mechanism, but they are directly used to support the ubiquitous computing system is inappropriate, because it is inappropriate to use objects to abstract components in ubiquitous computing.

As shown in Figure 1, we use component containers (Vessel) to encapsulate components. The component container provides the runtime environment for the components encapsulated in it. The component runs in this environment, and all its communication is carried out through the component container. Once a component is loaded into a container, it looks like a hard disk loaded into a hard disk box to the outside world, and only its external interfaces can be accessed, and the access to these interfaces must be done through the component container.

The runtime environment provided by the component container includes: activating the component when needed or passivating the component when it is not needed for a period of time, so as to protect limited system resources, such as main memory and CPU resources, etc.; Various service interfaces provide an adaptation layer, which avoids the trouble and potential safety hazards caused by client programs directly requesting services from components; provides a callback adaptation layer for components, and component containers can pass this callback The layer notifies components when interested events and messages occur; provides some configurable component parameters for clients to dynamically modify component properties.

The component container provides a component monitor for the components contained in it to complete the monitoring of the component lifecycle strategy. It is responsible for including: how to create components, whether to create permanent components or temporary components, when to activate components, when to passivate components, and when to destroy components; in addition, customers can also inform through the interface provided by the monitor Component Containers Instances of each component are no longer required.

The components produced by different manufacturers are different, and the components are packaged into a black box, and it is impossible for the outside world to know its internal implementation. However, before the client obtains the component, he must obtain enough information about the component in order to access it more effectively. For this reason, we introduce component control block CCB (Component Control Block) to provide and manage these useful information.

As shown in Figure 1, there is a one-to-one correspondence between component control blocks and components. Whenever the system creates a component, it will set a component control block, and then use the component control block to manage and control the component; , the system retracts its component control block, and the component dies. Therefore, the component control block is the only sign of the existence of the component. The role of component control block in component management is equivalent to that of process control block PCB in process management.

A component is encapsulated together with its control block in a component container. As shown in Figure 2, the component control block records a lot of meta information about the component, which can be divided into component description information and control information, including: component identifier, component attribute, component owner (someone in the network ubiquitous computing equipment), component creation time, component life, component granularity (class level, service level or function module level), component storage information, component function description, and component interface description. We bundle the component and its control block together, so that no matter what granularity or function the component can be managed conveniently.

It is worth noting that the description of meta-information by component control block must use widely used annotation language and modeling language (namely XML and UML), which can make component meta-information recognizable to most systems. We don't even need to know the granularity of the component in advance, but only need to read the specific information of the component from the component control block to check whether the component meets the needs of users, whether the component has shared attributes, and whether the storage space can accommodate the component. problem, and then access the component according to the description of the interface in the component control block. At the same time, it is also necessary to establish a system management configuration component library that is adaptive to the service environment, to complete the stress characteristics and dynamic reconstruction requirements of the component-based embedded system service environment, and to meet the individual requirements of different applications with a unified component interface and component service mode. The specific details of combining components into services are shielded in the service environment, and basic componentized application parts are provided to provide componentized services with individuality and dynamic stress characteristics that can adapt to the environment and needs.

Component containers, component monitors and component control blocks well solve component encapsulation, component life cycle management, and heterogeneous problems caused by non-uniform component granularity. Next, we use the component register to solve the problem caused by the high mobility of components.

Ubiquitous computing puts forward new definitions and requirements for the computing environment. People's understanding of the computing environment will change from a personal computer to a physical space with a variety of computing devices. We regard the entire physical space, rather than a single device in it, as a computing system; it needs to shield the details of each device in the physical space, and provide a standard program interface upwards, making this physical space a perceivable space. The operating system we design is to manage the resources of this physical space, so that applications can interact with the physical space. We call the objectively existing, organized, and expandable space containing certain computing capabilities physical space; the space formed by all the accessible components of each ubiquitous computing device is called the perception space Aware Space, as shown in Figure 4 Show.

Various computing devices in ubiquitous computing often communicate with each other through wireless means (such as infrared and bluetooth technologies), which is largely limited by the communication distance. Any device has a certain awareness space (Aware Space), as shown in Figure 4; the device cannot perceive components outside the awareness space. To this end, we introduce a component register CR (Component Recorder) to record all components that a certain device can perceive. Component-based services that support network distributed environments can make full use of network resources and reuse codes to obtain flexible component-based services that can be dynamically combined, and realize more complex functions with high efficiency.

As shown in Figure 3, the component register periodically detects components in the perception space, always keeping an up-to-date record of available components. Whenever a new component enters the perception space for a period of time, it will be detected; and a record information about the new component will be added in the component register. In the same way, if a registered component leaves the perception space, the component registerer will delete its record after periodic detection and finds that the component no longer exists.

In order to manage components more effectively, the record of the component register is divided into two parts. The first part records the local components owned by the device, and the second part records the components of other devices in the sensing space. Locally owned components are relatively stable and managed by the device itself; while network components change frequently, and the device only has the right to use them but not the right to manage them. When the system is initialized, all the records of the local components are firstly stored in the component register; then all network components in the sensing space are detected at that time, and their records are also added into the component register. When the system is running, the addition or deletion of local component records is determined by the device, which is very easy to manage; while network components need to be regularly checked for existence and availability.

Network components have three states: ready state, active state and dead state. Component dynamic registration adopts the lease method. Once the system finds a network component, it determines a lease period. When the lease period ends, the component registerer judges whether the network component is still available, thereby determining the next lease period. The component is in the ready state, and it is ready to be used by the system at any time; when the system needs to call the network component to complete a certain task, the component changes from the ready state to the active state, and the component starts the real work; after the component completes the task, It will return to the ready state; when the network component is deleted, or leaves the space that the system can perceive, or the system does not have sufficient permissions to access it, the component is in a dead state, and the dead component cannot be called by the system. Network components occupy system resources only when they are active.

In order to complete a certain task, the system needs components that meet certain requirements (for example, the component is required to complete the function A AND the required storage space of the component is less than B megabytes AND the life value of the component is greater than C seconds). The system makes requirements to CR to find out whether there are components that can meet these requirements. If such a component does not exist, return a null value to the system; if such a component exists and is unique, read the complete information of the component control block, and access the component according to the interface requirements; if there is more than one such component, read all of them The complete information of component control blocks that meet the conditions, and then compare the performance and network cost of these components (the network cost of the local component is zero), so as to select the best component to visit.

It is worth noting that since our design adopts the idea of "everything is a component", the component register is also a component, but it is rather special. It is automatically generated when the system is initialized. It is A component that the system must have. This design greatly facilitates the management of components, because there is nothing else but components.

Example:

Parse MPEG4 encoded files and play video streams. In this example, there are two computing devices: a personal computer running on the Linux operating system, and a PDA running on RT Linux; both devices are equipped with Bluetooth interfaces, and can be communication between each other. This example designs a total of four components: a component for parsing MPEG4, a component for displaying video, a component for parsing JPEG, and a component for displaying images; at the beginning, these four components were all running on a personal computer.

In the first step, the above four components are initialized on the personal computer. The initialization process of the component parsing MPEG4 is as follows: First, apply for a component control block and write various meta-information of the component into its component control block, as shown in Figure 2; then, use the component monitor to manage the components Life cycle strategy; next, the component and its control block and monitor are encapsulated into the component container; finally, the component container provides the running environment of the component, and all accesses must be carried out through the component container. The initialization process of the other three components is also basically the same.

In the second step, we gradually brought the PDA closer to the personal computer. Within a certain distance, PDA and personal computer can communicate through the Bluetooth interface. At this time, the PDA detects the four components on the personal computer, and stores their records on the component register. The component is in a ready state at this time, but does not occupy the storage space and computing resources of the PDA.

In the third step, the PDA needs to decode the MPEG4 file and play the file. At this time, search the components in the component register to check whether there is a component that can meet the requirements. Since in the second step, the component for parsing MPEG4 and the component for displaying video have been registered on the component register, so these two components can be easily found. Then, obtain sufficient information by accessing the component control blocks of the two components: including the function of the component, the interface of the component, and the size of the space occupied by the component. Finally, the PDA can use these two components to complete the decoding and playback tasks. When executing a task, the component changes from the ready state to the active state, which will occupy the storage space and computing resources of the PDA, as shown in Figure 5.

In the fourth step, after the MPEG4 file is played, the component for parsing MPEG4 and the component for displaying video will give up system resources and return to the ready state. Next, if we let the PDA stay away from the personal computer again, beyond the scope of the bluetooth effect, the above four components are no longer available, will be in a dead state, and also do not occupy the storage space and computing resources of the PDA.

Claims (8)

1. A component management method supporting pervasive computing, characterized in that the steps of the method are as follows: 1) In addition to the encapsulation of the component itself, the second encapsulation of the component is carried out through the component container, and the runtime environment of the component is provided. The so-called secondary encapsulation of the component through the component container: the component container provides The environment required for runtime, the component runs in this environment, and all its communication is carried out through the component container; once the component is loaded into the container, the client can only access the component through the external interface of the component container ; In the component container, a component monitor is also provided to complete the monitoring of the component life cycle strategy; 2) Manage the meta-information of the components at run time through the component control block, which includes component description information and control information. Each component is uniquely corresponding to a component control block, which is a concentrated reflection of component dynamics and is the system’s identification of components. and the basis of control, the so-called component control block is used to manage the meta-information of the component when it is running, which contains component description information and control information: the component control block and the component have a one-to-one correspondence relationship. Whenever the system creates a component, it will Set a control block for the component, and use it to manage and control the component; similarly, when the component will be revoked, the system will take back the component control block, and the component will also die; component control is the only sign of the existence of the component; the component control block The role played by the process control block in the component management is equivalent to that of the process control block; the description of the component information adopts the widely used annotation language and modeling language, namely XML and UML, which makes the component information relatively large Most systems are identifiable; 3) A component register is introduced to record all components that computing devices can perceive, and realize dynamic registration, replacement, addition and deletion of components. 2. A component management method supporting ubiquitous computing according to claim 1, characterized in that said component register records all components that computing equipment can perceive: always keep the latest record of available components; After a new component enters the perception space of the system for a period of time, it will be detected; and a record information about the new component will be added to the register; If you leave, the Registrar will delete its record after periodic detection and find that the component no longer exists. 3. A component management method supporting ubiquitous computing according to claim 1, characterized in that the environment required at runtime includes: 1) activating the component when needed or deactivating it when not needed for a period of time 2) Provide an adaptation layer for various service interfaces of commonly used components, which avoids the problems caused by client programs directly requesting services from components 3) Provide a callback adaptation layer for components, and the component container will notify components when interested events and messages occur through this callback layer; 4) Provide configurable component parameters for customers, use to dynamically modify the properties of components. 4. A component management method supporting ubiquitous computing according to claim 1, characterized in that said component monitor completes the monitoring of component life cycle policies, including: in what way to create components, to create permanent The component is still a temporary component, when the component needs to be activated, when the component needs to be passivated, and when the component needs to be destroyed; in addition, the client can use the interface provided by the monitor to inform that the instance of the component in the component container is no longer needed. 5. A component management method supporting ubiquitous computing according to claim 1, characterized in that said component description information and control information include: component identifier, component attribute, component owner or Computing equipment in the network, the creation time of components, the life of components, the granularity of components, that is, object level, service level, or function module level, component storage information, component function description, and component interface description. It is bundled with its control block, so that no matter what granularity or function, components can be easily managed. 6. A component management method supporting pervasive computing according to claim 1, characterized in that the description of said component meta-information adopts widely used annotation language and modeling language, namely XML and UML: Make component information identifiable to most systems; through this information, we don't even need to know the granularity of the component in advance, but only need to read the specific information of the component from the component control block to check whether the component meets the needs , whether the component has shared attributes, whether the storage space can accommodate the component, and then access the component according to the description of the component interface in the control block. 7. A component management method supporting ubiquitous computing according to claim 2, characterized in that the latest record of said components is divided into two parts: the first part records the local components owned by the computing device, and the second part Record the components of other computing devices in the sensing space; the locally owned components are relatively stable and managed by the computing device itself, while the network components have three states and change frequently, and the computing device only has the right to use it but not the right to manage it; in the system When initializing, first store all records of local components in the component register, then detect all network components in the system at that time, and add their records to the component register; when the system is running, the addition of local component records or Deletion is determined by the computing device, which is very easy to manage, while network components need to be periodically checked for existence and availability. 8. A component management method supporting ubiquitous computing according to claim 2 or 7, characterized in that said network components have three states: ready state, active state and dead state; the dynamic registration of components adopts the way of leasing , once the system finds a shared component in the network, it determines a lease period. When the lease period ends, the component registerer judges whether the network component is still available, thereby determining the next lease period. The component during the lease period is in a ready state. It is ready to be used by the system at any time; when the system needs to call the network component to complete a certain task, the component will change from the ready state to the active state, and the component will start the real work; after the component completes the task, it will return to the ready state State; when a network component is deleted, or leaves the space that the system can perceive, or the system does not have sufficient permissions to access it, the component is in a dead state, and the dead component cannot be called by the system.

Images