GB/T 41780.2-2024Internet of things - Edge computing - Part 2: Data management requirements (English PDF)

3 Terms and definitions

Six terms are defined. Edge computing (3.1) is a computing architecture that places data processing and storage on the edge side of the internet of things, taken from GB/T 41780.1-2022, 3.1. An edge computing node (3.2) is a device on the edge side with data processing, network communication and storage capability, a note naming internet of things terminals, edge gateways, edge controllers and edge computing servers, taken from GB/T 41780.1-2022, 3.2. A data management function (3.3) is a function of an edge computing node relating to the acquisition, aggregation, processing, storage, transmission, sharing, display, deletion and security of data.

Data description (3.4) is the representation of data in a suitable form on the basis of its features and regularities, providing the input for data modelling. Data quality (3.5) is the degree to which the characteristics of data satisfy stated and implied needs when used under specified conditions, taken from GB/T 36344-2018, 2.3. Metadata (3.6) is data that defines and describes other data, taken from GB/T 18391.1-2009, 3.2.16. The abbreviations of Clause 4 are CoAP, HTTP, MQTT, OPC UA, TCP and UDP.

5 Functional framework for edge computing data management

5.1 Following 5.1 of GB/T 41780.1-2022 on the system architecture, the kinds of edge computing node are the internet of things terminal, the edge gateway, the edge controller and the edge computing server. Following the functional architecture of the same subclause, the functional framework for data management is shown in Figure 1: the data management functions of each kind of node are data acquisition, data aggregation, data processing, data storage, data transmission, data sharing, data display, data deletion and data security, and the functions may be selected as the deployment needs. Every kind of edge computing node shall have the data security function. The cloud, or data centre, of Figure 1 may be a private cloud, a public cloud, a hybrid cloud or a data centre, and the legend of the figure distinguishes edge computing points from cloud or data centre and mandatory from optional functions.

5.2 Table 1 describes the data management functions of the four kinds of node in a five column matrix, one column for the function and one for each kind of node, and marks with a dash the combinations that do not apply. Data acquisition is carried out by the internet of things terminal alone and is described as converting the information of the target object, such as a thing or an environment, into a data description according to a certain regularity by means of a sensing unit. Data aggregation is carried out by the gateway, the controller and the server, and brings distributed, cross network, multi source data in over several data interfaces. Data sharing is carried out by the edge computing server alone and provides data exchange with external systems and open sharing.

Table 1 continues with the functions that all four kinds of node share. Data processing is described for the terminal as pre-processing data according to certain rules, or converting control instruction data into control execution according to a certain regularity, the printed Chinese phrase for what the instruction data is converted into reading control execution force; for the gateway it is protocol conversion together with analysis, filtering, coding and compression of the data and parsing and issuing of control instructions; for the controller it is complex logic control, parameter control optimization, control protocol conversion and data response; and for the server it is complex data processing and analysis providing edge services. Data storage is temporary caching on the terminal; short term storage of acquired data and long term storage of system configuration data on the gateway, with local and expansion storage and storage monitoring and data backup; and, on the controller and the server, storage management, verification updating and redundant backup on top of what the gateway does.

Table 1 ends with data transmission, supported on the terminal for exchange with an upper level edge computing node or the cloud, and on the other nodes for the same exchange plus, where necessary, exchange between nodes of the same level; with data display, for which an edge computing node should provide a real time visual data interface; with data deletion, where the storage resources of an edge computing node are limited and the acquired data that has passed its storage period is to be deleted automatically according to the needs of the application scenario; and with data security, where data is classified and graded for protection according to its importance and the degree of harm that its tampering, destruction or leakage would cause.

6 Data acquisition requirements

The data acquisition of the internet of things terminal shall guarantee the precision, the period and the quality of the acquisition; shall meet the requirements of the system for regularity, completeness, accuracy, consistency and timeliness of the data; shall not affect the normal running of the business of the edge computing node; and shall include the identifier of the terminal in the acquired data. It should support multi-channel acquisition, should support marking the first and the last position of the acquired data, and should timestamp the acquired data.

7 Data aggregation requirements

7.1 Data aggregation shall be able to obtain dynamic or non-dynamic data from multiple data sources, such as the target objects and the underlying devices, over different data interfaces; shall be able to recognize the source of the data; shall be able to recognize the type of the data; and shall include the identifier of the edge computing node in the aggregated data. It should support offline or real time acquisition and should provide visual configuration, management or monitoring tools that make configuration and management more efficient and convenient.

7.2 The edge gateway shall be able to obtain data concurrently over several data interfaces, should support structured, semi-structured and unstructured data, and should be able to obtain data relating to the edge computing network communication. 7.3 The edge controller shall be able to obtain real time data and control command feedback data, should be able to obtain analogue data, and should be able to obtain data concurrently over several data interfaces. 7.4 The edge computing server shall support structured, semi-structured and unstructured data and concurrent acquisition over several data interfaces, and should support the publish and subscribe acquisition mode.

8 Data processing requirements

8.1 Data processing shall be able to split and reassemble data packets, should be able to pre-process data by cleaning and extraction, should be able to load data processing modules dynamically so as to suit different kinds of data, and should be able to configure timed data analysis.

8.2 The internet of things terminal shall support converting the acquired data into a digital signal and, where it has an actuating device, shall guarantee the timeliness and the accuracy of the response to control instruction data. It should support event based acquisition, a configurable processing mode, conversion of the acquired data into a unified data description, and selection of the acquired data by particular rules; it should not carry out complex data analysis, so that computing resources are not occupied for a long time.

8.3 The edge gateway shall be able to filter and decode data, and should support interpreting and converting data of different protocols into a unified data model, extracting features, coding, packing and compressing the data, managing the issue of simple control instructions, a data rule engine and statistical analysis; it may have a data quality assessment capability that improves the value of the data in use. 8.4 The edge controller shall support data processing for complex logic control, shall be able to respond in good time to process control and emergency event data, and shall guarantee the timeliness of the processing; it should support rule engine configuration for complex business logic, data aggregation and fusion, optimization of control system parameters and event based data analysis.

8.5 The edge computing server shall be able to recognize identifiers automatically and parse data labels and ciphertext, shall classify, reassemble and fuse the data obtained by the edge computing nodes, shall support a data rule engine and shall support several forms of output of the analysis results. It should support elastic invocation of data computing resources, statistical analysis and classification recognition and other edge analysis functions, should provide a big data analysis framework and modules, should build the algorithm library needed for data mining and analysis, and may have a data quality assessment capability.

9 Data storage requirements

9.1 Data storage shall be able to cache and overwrite data and shall fix the storage period of the different types of data according to the needs of the application scenario; inside an edge computing node the storage mode and location should be chosen according to the type, the features and the importance of the data, storage should be able to be visualized, and data should be stored on different edge computing nodes according to how often it is accessed. 9.2 The internet of things terminal should set an expiry time on the stored data so that it stays current, and should compress the stored data.

9.3 The edge gateway shall support local historical data storage that is not lost when the power fails, shall be able to expand storage externally, and shall provide automatic or manual data backup; it should be able to create data identifiers, to retrieve data, to monitor storage resources and data information, to update and delete data in batches, and should store data in a database. 9.4 The edge controller shall support local historical data storage that survives a power failure, automatic or manual backup and real time data storage, and should be able to create data identifiers and to monitor storage resources and data information.

9.5 The edge computing server shall create traceability identifiers for the data, shall create metadata describing the features and attributes of the data, shall index the data and be able to retrieve it, shall support the storage and management of real time and historical data and of several different data types, and shall keep the stored data through a power failure. It should provide metadata creation and management tools, should support a data dictionary, should verify the stored data at regular intervals, should update the storage resources and delete expired data at regular intervals, and should support dynamic expansion of the storage resources.

10 Data transmission requirements

10.1 Data transmission shall provide the data transmission interfaces between edge computing nodes together with their interface descriptions, shall choose a suitable transmission mode according to the timeliness required, and shall be able to continue or repeat a transmission that has been interrupted; edge computing nodes of the same level may transmit data between themselves as the application needs. 10.2 The data transmission of the internet of things terminal should keep the transmission resources it occupies small and should not affect the running of other business.

10.3, 10.4 and 10.5 each require the node to recognize different network communication protocols and to support one or more of TCP, UDP, OPC UA, MQTT, CoAP and HTTP. In addition the edge gateway shall be able to transmit data concurrently with several edge computing nodes; the edge controller should be able to transmit data with other edge controllers or edge gateways; and the edge computing server shall be able to transmit data concurrently with several edge computing nodes and shall improve the response speed of concurrent data access by means of thread pools, dynamic resource allocation and similar techniques.

11 Data sharing requirements

The data sharing function of the edge computing server shall follow the principle of the smallest sharing scope and the smallest shared data set, shall provide standard data interfaces together with their descriptions, shall keep the data that is to be shared accessible, and shall be able to exchange data concurrently so that it can share data with several systems at the same time; it should be flexible enough to support the exchange protocols and data formats of different systems.

12 Data deletion requirements

The data deletion function of an edge computing node shall configure the storage period and the deletion policy of the different types of data according to the needs of the application scenario; it should classify and grade the data and choose a destruction method suited to each class and grade, including but not limited to logical destruction, encryption destruction and physical destruction; and it may supervise and audit the deletion process so that deletion stays traceable.

13 Data security requirements

13.1 The document sets out the data security requirements for the internet of things terminal, the edge gateway, the edge controller and the edge computing server, and the basic security protection and general security requirements of the internet of things system shall meet GB/T 37044-2018.

13.2 The data security of an edge computing node shall include a unique identifier when the node joins the network, the inclusion of that unique identifier in the data uploaded, and access control over the data on the least privilege principle. The node should draw up data classification rules and store and manage the data by class and grade; should be able to desensitize sensitive data; should store data encrypted; should decide whether to encrypt transmission according to the type of data, so that the confidentiality and integrity of the transmission are protected; should establish a maintenance system and maintenance process for the data with defined technical measures against tampering and leakage; and should log user operations and system events and keep those records in a secure place for audit.