CN109245933A

CN109245933A - One kind calculating environmental resource towards cloud combination of edge and finds method

Info

Publication number: CN109245933A
Application number: CN201811105125.5A
Authority: CN
Inventors: 高岭; 杨旭东; 郭洪波; 冯通; 王海; 郑杰; 张晓�
Original assignee: Northwest University
Current assignee: Northwest University
Priority date: 2018-09-21
Filing date: 2018-09-21
Publication date: 2019-01-18

Abstract

A method for resource discovery in a cloud-oriented converged edge computing environment, completes resource discovery through three main steps: resource discovery, resource description, and resource service interface, solves the problem of difficulty in discovering massive heterogeneous resources in the edge computing environment, and provides resource Scheduling and task migration lay a good foundation.

Description

One kind calculating environmental resource towards cloud combination of edge and finds method

Technical field

The invention belongs to edge calculations fields, and in particular to a kind of to calculate environmental resource discovery side towards cloud combination of edge Method.

Background technique

The appearance of cloud computing solves the problems, such as that mobile terminal operational capability is insufficient, and the calculation power of mobile terminal is promoted also certain Degree solves the problems, such as that cloud computing spends time delay high, but solves cloud computing time delay height simultaneously to calculate power deficiency with terminal to be current The huge challenge faced.Simultaneously in application field, the prior art has no ability to cope in reply complex scene and emergency event. But protect and monitor in platform, it is limited by that monitoring side ability is limited, transmission network building expense in field is big, low efficiency and real When monitoring and accurate operational tupe, it is difficult to complete realize carries out comprehensive place immediately to the Great Wall monitoring information of acquisition Reason, it is difficult to accurately collected data Accurate Analysis be modeled, cause processing delay high, operational precision is low to be difficult to meet Great Wall The comprehensive monitoring of ruins perception and the demand of accurate risk early warning.I.e. existing Ruins of Great Wall perceptual model is difficult to complex environment Under real-time task provide quick response, accurate operational service edge calculating come into being.Edge calculations first have to calculating The resources such as terminal, data, cloud service platform in system or environment are found, while how to be provided service after finding and be also The great difficulty that edge calculations face.

Summary of the invention

In order to overcome the above-mentioned deficiencies of the prior art, the object of the present invention is to provide one kind calculates towards cloud combination of edge Environmental resource finds method, difficult in order to solve resource low delay discovery in edge calculations environment；The description of resource High Availabitity is difficult to ask Topic is resource for Unify legislations such as equipment, data, services in edge calculations environment.According to the demand of edge calculations environment, The discovery of resource is particularly important as the basis of the offer service of edge calculations.

To achieve the goals above, the technical solution adopted by the present invention is that:

One kind calculating environmental resource towards cloud combination of edge and finds method, comprising the following steps:

1) discovery of equipment and data resource:

It is mainly used for calculating required data and is found with terminal, magnanimity, isomery, poor arithmetic ability is dispersed in peripheral surroundings How different biggish physical equipment finds that these device resources are very crucial, based on actively discover with by the discovery machine of dynamic response System, can effective shielding device structure and operational capability difference, all devices in comprehensive discovery domain in time；The active of equipment It was found that with two aspects of passive finding:

Equipment actively discovers:

Active resource discovery method based on point-to-point model, in the case that Resource Server operates normally, by Resource Server As server end, equipment is as client, after the completion of deployment, establishes the discovery that resource is realized in communication by application, simultaneously Server-side is requested using real-time audiomonitor, completes facility registration；

When Resource Server missing or low-response are difficult to meet the needs of service, equipment actively proposes facility registration request, to It succeeds in registration after Resource Server response, completes actively discovering for equipment, point-to-point services model does not distinguish client and service The software application at end, unified application program, the i.e. role of equipment and server are opposite rather than fixed, the ginsengs in system A part of hardware resource that they are possessed, i.e. processing capacity, storage capacity, network-connectivity, printer are shared with person, this A little shared resources can directly be accessed without being subjected to intermediate real by network offering services and content by other peer node Peer Body, the participant in this network are both the supplier Server and resource, service and content of resource, service and content Obtain taker Client；

The passive finding of equipment:

Passive resource discovery method based on CLIENT/SERVER MODEL, using Resource Server as server end, equipment is as visitor The discovery that resource is realized in communication is established by application, while server-side is asked using real-time audiomonitor after the completion of deployment in family end It asks, completes facility registration, application software is divided into two parts by CLIENT/SERVER MODEL:

A part is client software, it is operated in user's local；

Another part is server software, it is operated on the server host on network；

Client and server belong to the application software layer of TCP/IP, they transmit data using TCP or udp protocol, and It is communicated in accordance with certain application protocol, realizes specific network application function jointly, the interactive mode of client and server is: I.e. continuous service, monitoring wait the service request of client on its serve port after service processes starting, and user uses client's journey Sequence is interacted with service processes, and after client process starting, system is that it distributes a transient port, is used for receiving data, client As some serve port sending service request of server host, asking for service processes customer in response on that port is monitored It asks, completes specified operation；Then by the Client IP address and source port number of band in request data package, visitor is returned result to Family program；Such a complete interactive process is known as a session；Client and clothes are established by client and server application Business device end, while being communicated using the conversation mechanism of application layer, realize comprehensive discovery of resource；

2) unification of resource accurately describes:

Resource after being mainly used for discovery carries out the foundation of resource table, carries out resource simultaneously for the resource for being newly added or leaving The update of table；

The unification of resource accurately describes:

The foundation of resource table: how the crucial resource primarily directed to magnanimity isomery that resource table is established carries out retouching for standard unification It states, is a kind of resource description standard of the invention, edge calculations system distributes network address to each equipment, and each resource possesses one A to describe resource type, the name of the current information of Resource Properties and element of resource, wherein resource type is by preceding 5 types Position, 5-15 bit attribute position and chained list element position are constituted, and the name of each resource is unique and identifies for service management, equipment And part replacement；

The update of resource table: edge calculations system distributes network address to each equipment, and each equipment of internal system possesses one A to describe position, allocation identification is met network address and distributes to edge by the name of role and current information, edge network Equipment, the name of each equipment is unique and is used for service management, equipment identification and part replacement, and the name of equipment can be correct Be matched to device identifier and network address, identifier is used for the equipment management of edge calculations system, and network address IP address or MAC Address, then for supporting such as 4G, NB-IoT, bluetooth, the various network communications agreement such as Zigee or WiFi；

More new range: under peripheral surroundings smaller in range and fixed, network is distributed to each equipment using edge calculations system Address, each equipment of internal system, which possesses one, can describe position, the name of role and current information, and edge network will divide Meet network address with mark and distribute to edge device, the name of each equipment is unique and knows for service management, equipment Other and part replacement.This naming rule makes user and ISP is easier to management service, while organization service provider obtains To hardware information, private data is protected, the name of equipment can correctly be matched to device identifier and network address, and identifier is used for side The equipment management of edge computing system, and network address IP address or MAC Address, then for supporting such as 4G, NB-IoT, bluetooth, The various network communications agreement such as Zigee or WiFi；

The fusion of cross-domain resource table: after resource table is established, according to the resource table between the edge domain of different Edge Servers establishment It needs to update, precisely be described to solve the resource between edge domain, propose a kind of resource description model for merging polygon marginal area；

3) it provides uniform services interface for the resource deployment and task immigration service on upper layer:

It is mainly used for providing reliable resource perception service for edge calculations top service, solves after resource table merges in order to upward Layer provides the service of response, improves the accuracy of resource description, and reduces data storage scheduling overhead and provide for resource distribution Service proposes accurately to describe method and resource map service based on the associated resource table data of non-significant Feature Semantics.

Further, the core procedure of equipment described in step 1) actively discovered is as follows;

1) server, deployed with devices client application；

2) equipment sends registration request:

If server-side agrees to registration；

Server sends response of succeeding in registration；

Dispose Client application；

The equipment is found；

Else server-side disagrees registration；

Server-side transmission rejects response；

The equipment is not found.

Further, the core procedure of the passive finding of equipment described in step 1) is as follows;

1) Resource Server broadcast resource discovery request；

2) deployment services device application simultaneously；

If equipment agrees to discovery；

Equipment sends unicast and agrees to response；

Dispose Client application；

Equipment end application sends registration request；

Server returns to response of succeeding in registration；

The equipment is found；

Else equipment disagrees discovery；

Equipment transmission rejects response；

The equipment is not found.

Further, the step of polygon marginal area of fusion described in step 2:

Firstly, the selection of description indexes and the expression of resource table:

1) how the index for constructing resource table is chosen: the associated selecting index of terminal mobile context;

2) how resource table indicates: the representation method of set, dictionary;

Secondly, using the selection of the Fusion Model Fusion Model based on general character context information:

1) similar context information filtering selection: how quality is carried out according to the context information that sound constructs, filters out repetition scene Information;

Core key is data filtering problem, is mainly taken based on the set of metadata of similar data selection of the similar entropy of sound multi objective and based on sound The data filtering method of sound spectrum quality;

2) cloud context information stores: according to the transformation of context information and environment, which storing data in, first predicts scene-again Fusion, Scenario Prediction, sound scene and the associated storage strategy of data information of space-time transformation select；

3) sensitive scene secret protection information fusion: according to the privacy of context information, how to be merged, core key is hidden Private and final energy limitation problem, it is mainly limited using user demand and the associated Fusion Model of context information and final energy Fusion Model is associated with scene；

Finally, fusion method is based primarily upon the fusion method of sound scene event code:

1) coding method of frequency spectrum entropy and turnover rate: entropy determines code length method, and slamp value determines code value method；

2) code length and code value: the determination method of code length, code value determine method；

3) maximum value matching the maximum matching Fusion Model of edge domain coding: is carried out according to the sound event after coding.

Further, method and money are accurately described based on non-significant Feature Semantics associated resource table data described in step 3) Source figure service, steps are as follows:

1) resource characterizes feature selecting: including perceiving the non-significant semantic feature selection method of sequential correlation and being based on similar language Adopted feature Enhancement Method；

2) accurate resource description: the multivariate resource of the concept graph code including geographic location association precisely describes method and scene- The multi-source of information binary order-preserving column knowledge mapping coded representation precisely describes method；

3) resource map service: including adaptive universe resource map service.

Further, the name of the current information of Resource Properties and element of resource described in step 2, edge calculations system are given Each equipment distributes network address, and each resource, which possesses one, can describe resource type, and Resource Properties and element of resource are worked as The name of preceding information.5-15 bit attribute position and chained list element position are constituted, and the name of each resource is unique and for servicing pipe Reason, equipment identification and part replacement.

Further, the universe resource map is suitble to the resource of user for the different demands of user in resource space It is not quite similar, therefore just needs to construct specific available resource set for different user demands, face use space system The resource of description, the actual demand of user need to have the mapping relations with resource space, could become Specific measurement index in resource space, measure resource availability when, according to user demand mapping measurement standard come The resource individual for judging various discrete in resource grid, to determine whether this resource unit can be used. So these according to user demand Mapping standard formed resource grid in available resource set Referred to as universe resource map.

Further, the universe resource map of the resource, constructs manageable basic resource unit, in airspace, time domain, meter It calculates, storage, the description of structure and characterizing method after all carrying out discretization in multiple dimensions such as data to resource space, in face of making The resource described with three-dimensional system, the actual demand of user need to have the mapping relations with resource space, could become and provide Specific measurement index in source space.

The beneficial effects of the present invention are: increasing substantially the accuracy rate of resource discovering in edge calculations environment, promote to calculate Resources effective utilization in environment；Unnecessary expense is reduced simultaneously, provides important leverage for edge calculations service.

Detailed description of the invention

Fig. 1 is resource perception technological frame figure of the present invention；

Fig. 2 is that present device resource actively discovers procedure chart；

Fig. 3 is present device resource passive finding procedure chart；

Fig. 4 is that edge inter-domain resource of the present invention collaboration updates frame diagram；

Fig. 5 is that resource of the present invention precisely describes service interface figure.

Specific embodiment

The present invention is further discussed below below in conjunction with attached drawing, but the present invention is not limited to following embodiment.

As shown in Figure 1, a kind of calculate environmental resource discovery method towards cloud combination of edge, comprising the following steps:

1) discovery of equipment and data resource:

As shown in Fig. 2, equipment actively discovers:

As shown in figure 3, the passive finding of equipment:

A part is client software, it is operated in user's local；

2) unification of resource accurately describes:

The unification of resource accurately describes:

As shown in figure 4, the fusion of cross-domain resource table: resource table establish after, according to different Edge Servers establish edge domain it Between resource table need to update, precisely described to solve the resource between edge domain, propose a kind of money for merging polygon marginal area Source Description model；

As shown in figure 5,3) provide uniform services interface for the resource deployment and task immigration service on upper layer:

1) server, deployed with devices client application；

2) equipment sends registration request:

If server-side agrees to registration；

Server sends response of succeeding in registration；

Dispose Client application；

The equipment is found；

Else server-side disagrees registration；

Server-side transmission rejects response；

The equipment is not found.

1) Resource Server broadcast resource discovery request；

2) deployment services device application simultaneously；

If equipment agrees to discovery；

Equipment sends unicast and agrees to response；

Dispose Client application；

Equipment end application sends registration request；

Server returns to response of succeeding in registration；

The equipment is found；

Else equipment disagrees discovery；

Equipment transmission rejects response；

The equipment is not found.

Further, the step of polygon marginal area of fusion described in step 2:

2) how resource table indicates: the representation method of set, dictionary;

3) resource map service: including adaptive universe resource map service.

Claims

1. a cloud-oriented fusion edge computing environment resource discovery method, is characterized in that, comprises the following steps:

1) Discovery of devices and data resources:

It is mainly used for data and terminal discovery required for computing. There are massive, heterogeneous physical devices with large differences in computing capabilities distributed in the edge environment. How to discover these device resources is very critical. Based on the discovery mechanism of active discovery and passive response, It can effectively shield the differences in device structure and computing power, and discover all devices in the domain in a timely and comprehensive manner. There are two aspects of active discovery and passive discovery of devices:

Active discovery of devices:

The active resource discovery method based on the point-to-point model, when the resource server is running normally, the resource server is used as the server and the device is used as the client. Request to complete device registration;

When the resource server is missing or the response is slow and it is difficult to meet the service requirements, the device takes the initiative to submit a device registration request. After the resource server responds, the registration succeeds, and the active discovery of the device is completed. The point-to-point service model does not distinguish between client and server software applications, and unified application programs. , that is, the roles of the device and the server are relative rather than fixed, and the participants in the system share a part of the hardware resources they own, that is, processing power, storage capacity, network connection capacity, and printers. These shared resources provide services and Content can be directly accessed by other peer nodes without going through an intermediate entity. Participants in this network are both the provider of resources, services and content Server, and the acquirer of resources, services and content Client;

Passive discovery of devices:

The passive resource discovery method based on the client/server model uses the resource server as the server and the device as the client. After the deployment is completed, the application establishes communication to realize the resource discovery, and the server application monitors the device request in real time and completes the device registration. The client/server model divides the application software into two parts:

Part of it is client software, which runs on the user's local machine;

The other part is the server-side software, which runs on the server host on the network;

Both the client and the server belong to the application software layer of TCP/IP. They use the TCP or UDP protocol to transmit data, and communicate according to a certain application protocol to jointly implement specific network application functions. The interaction between the client and the server is: The service process starts After that, it will continue to run, listen on its service port, and wait for the customer's service request. The user uses the client program to interact with the service process. After the client process starts, the system allocates a temporary port for it to receive data. The client is like the server host. A service port sends a service request, and the service process listening on the port responds to the client's request and completes the specified operation; then returns the result to the client program according to the client's IP address and source port number in the request packet; Such a complete interaction process is called a session; the client and the server are established through the client and server applications, and the session mechanism of the application layer is used for communication to realize the comprehensive discovery of resources;

2) A unified and precise description of resources:

It is mainly used to establish the resource table for the discovered resources, and to update the resource table for newly added or left resources;

A uniform and precise description of the resource:

Establishment of the resource table: The key to the establishment of the resource table is to describe how to standardize and describe the massive heterogeneous resources. To this end, a resource description standard is invented. The edge computing system assigns a network address to each device, and each resource has a It can describe the naming of resource types, resource attributes and current information of resource elements. The resource type consists of the first 5 type bits, 5-15 attribute bits and linked list element bits. The naming of each resource is unique and used for services. management, equipment identification and parts replacement;

Update of the resource table: The edge computing system assigns a network address to each device. Each device in the system has a name that can describe the location, role, and current information. The edge network assigns an identifier that matches the network address to the edge device. The naming of the device is unique and used for service management, device identification and component replacement. The naming of the device can be correctly matched to the device identifier and the network address. The identifier is used for device management of the edge computing system, and the network address IP address or MAC address. The address is used to support various network communication protocols such as 4G, NB-IoT, Bluetooth, Zigee or WiFi;

Update scope: In a small and fixed edge environment, the edge computing system is used to assign a network address to each device. Each device in the system has a name that can describe the location, role and current information, and the edge network will assign an identifier. Conforms to the network address assigned to edge devices, and the naming of each device is unique and used for service management, device identification, and component replacement. This naming rule makes it easier for users and service providers to manage services, while organizing service providers to get hardware information , protect privacy data, the naming of the device can correctly match the device identifier and network address, the identifier is used for device management of the edge computing system, and the network address IP address or MAC address is used to support devices such as 4G, NB-IoT, Various network communication protocols such as Bluetooth, Zigee or WiFi;

Fusion of cross-domain resource tables: After the resource table is established, the resource table between edge domains established by different edge servers needs to be updated. In order to solve the accurate description of resources between edge domains, a resource description model that integrates multiple edge domains is proposed. ;

3) Provide a unified service interface for the upper-layer resource deployment and task migration services:

It is mainly used to provide reliable resource-aware services for the upper-layer services of edge computing, solve the service that provides response to the upper layer after the fusion of resource tables, improve the accuracy of resource description, and reduce the data storage scheduling overhead to provide services for resource configuration. Resource table data accurate description method and resource graph service of salient feature semantic association.

2. A cloud-oriented fusion edge computing environment resource discovery method according to claim 1, characterized in that, the core steps of active discovery of the device in step 1) are as follows;

1) Server, device deployment client application;

2) The device sends a registration request:

if the server agrees to register;

The server sends a registration success response;

Deploy client applications;

the device was discovered;

else the server does not agree to the registration;

The server sends a rejection response;

The device is not found.

3. A cloud-oriented fusion edge computing environment resource discovery method according to claim 1, characterized in that, the core steps of passive discovery of the device in step 1) are as follows;

1) The resource server broadcasts a resource discovery request;

2) Deploy server applications at the same time;

if the device agrees to discover;

The device sends a unicast consent response;

Deploy client applications;

The device-side application sends a registration request;

The server returns a successful registration response;

the device was discovered;

else the device disagrees with discovery;

The device sends a rejection response;

The device is not found.

4. A cloud-oriented fusion edge computing environment resource discovery method according to claim 1, characterized in that, the step of merging multiple edge domains described in step 2):

First, describe the selection of indicators and the representation of the resource table:

1) How to select the indicators for constructing the resource table: the selection of indicators associated with terminal mobile scenarios;

2) How to represent the resource table: the representation of collections and dictionaries;

Secondly, the selection of fusion model based on the fusion model based on common scenario information is adopted:

1) Similar scene information filtering selection: how to quality and filter out the repeated scene information according to the scene information constructed by the sound;

The core key is the data filtering problem, which mainly adopts the similar data selection based on sound multi-index similarity entropy and the data filtering method based on sound spectrum quality;

2) Cloud context information storage: According to the transformation of context information and the environment, where to store the data, first predict the context - and then integrate, the context prediction of spatio-temporal transformation, and the storage strategy selection of the association between sound context and data information;

3) Sensitive context privacy protection information fusion: According to the privacy of context information, how to integrate the core key is the issue of privacy and terminal energy limitation. Association fusion model;

Finally, the fusion method is mainly based on the fusion method of sound scene event coding:

1) The encoding method of spectral entropy and flow rate: the entropy value determines the encoding length method, and the flow value determines the code value method;

2) Code length and code value: the method of determining the length of the code and the method of determining the code value;

3) Maximum matching fusion model of edge domain coding: maximum matching is performed according to the encoded sound events.

5. A cloud-oriented fusion edge computing environment resource discovery method according to claim 1, characterized in that, in step 3) the method for accurately describing resource table data based on non-salient feature semantic association and the resource graph service, step 3) as follows:

1) Resource representation feature selection: including non-salient semantic feature selection methods for perceiving temporal associations and enhancement methods based on similar semantic features;

2) Precise resource description: a multi-source accurate description method including the concept map coding of geographic location associations and a multi-source accurate description method represented by the context-information binary sequence-preserving knowledge map coding;

3) Resource graph service: including adaptive global resource graph service.

6. A method for discovering resources in a cloud-oriented converged edge computing environment according to claim 1, characterized in that, for the naming of the resource attributes and the current information of resource elements in step 2), the edge computing system assigns each device Network address, each resource has a name that can describe the resource type, resource attributes and current information of resource elements, composed of 5-15 attribute bits and linked list element bits, the name of each resource is unique and used for service management, Equipment identification and parts replacement.

7. A method for discovering cloud-oriented converged edge computing environment resources according to claim 5, characterized in that, for the global resource map, according to the different needs of users, the resources suitable for users in the resource space are not the same, Therefore, it is necessary to build a specific set of available resources for different user needs. Faced with the resources described by the space system, the actual needs of users need to have a mapping relationship with the resource space in order to become a specific measurement index in the resource space. When the availability of resources, each discrete resource individual in the resource grid is judged according to the measurement standard of the user demand mapping, so as to determine whether the resource unit is available, then the available resource sets in the resource grid formed according to the user demand mapping standard , which can be called the global resource graph.

8. The method for discovering resources in a cloud-oriented converged edge computing environment according to claim 7, wherein the global resource map of the resources constructs a manageable basic resource unit, in the air domain, time domain, computing, In terms of storage, data and other dimensions, the structure and representation method of the resource space after discretization are described. In the face of resources described by the space system, the actual needs of users need to have a mapping relationship with the resource space in order to become a resource space. Specific metrics in the space.