CN114666682A

CN114666682A - Multi-sensor Internet of things resource self-adaptive deployment management and control middleware

Info

Publication number: CN114666682A
Application number: CN202210298276.7A
Authority: CN
Inventors: 陈同中
Original assignee: Individual
Current assignee: Individual
Priority date: 2022-03-25
Filing date: 2022-03-25
Publication date: 2022-06-24

Abstract

The method is designed and constructed for self-adaptive deployment and control middleware of multi-sensor Internet of things resources, and dynamic decision of data is carried out based on environmental proximity calculation; the middleware is designed to realize the convergence access deployment of various wireless communication network nodes; through virtualization of sensor node attributes and processing in a sensing network, a virtual sensor abstract pool is formed inside a middleware platform, implementation details of heterogeneous sensing nodes and communication protocols are shielded, a real-time dynamic control message publishing and subscribing process of an application end in aspects of node resource discovery and deletion, node running state monitoring, node working modes and the like is designed, and end-to-end interaction between sensor nodes and the application end is realized; data filtering and calculation are added into the middleware, interconnection and intercommunication among different devices of the sensor network are achieved, the problem that heterogeneous network information nodes are difficult to fuse and transmit is solved, sensing layer devices and data types are diversified, and the requirements of various industries in the future are met well.

Description

Multi-sensor Internet of things resource self-adaptive deployment management and control middleware

Technical Field

The application relates to a management and control middleware for deployment of a sensor Internet of things, in particular to a management and control middleware for self-adaptive deployment of multi-sensor Internet of things resources, and belongs to the technical field of the middleware of the Internet of things.

Background

With the development of wireless sensor networks and mobile internet and the emergence of cloud computing technology, the internet of things has been widely deployed in various applications such as environmental monitoring, smart home, smart agriculture, urban traffic management and control, and the like. The intelligent sensing and digital identity integrated intelligent service system integrates intelligent sensing, digital identity, ubiquitous network and cloud computing technologies, connects the terminal sensing equipment with the Internet, and achieves comprehensive sensing and intelligent service. The internet of things comprises embedded electronic equipment, software, sensors and a communication network, the objects can collect, forward and exchange data, remote network facilities can interactively manage the objects, and each object can be uniquely identified and matched with each other.

The internet of things is constructed by the following layers: (1) a sensing layer: including sensing devices and communication networks formed between devices. (2) Network layer: various environmental data collected by the sensing layer are transmitted to the application layer through various network infrastructures, and the application layer is a bridge for communication between the local sensor network and the wide-area mobile communication network and the internet. (3) An application layer: the sensing data is processed and analyzed by means of technologies such as cloud computing and the like, services meeting various industry requirements are provided, the industry intellectualization is realized, and meanwhile, the system is also a decision layer and is used for distributing and controlling sensing tasks of the lower two layers according to the requirement change.

In order to achieve the purposes of comprehensive observation and intelligent perception, the internet of things needs to absorb and fuse various technical advantages in the information field. Under the environment of a single perception network, the communication distance, the transmission bandwidth and the network topology are relatively fixed, the perception of different types of data cannot be simultaneously met, the application environment conditions are complex and changeable, the types and the data volume of the collected data are continuously increased, and the current application requirements cannot be met only by deploying the single perception network.

The deployment of the heterogeneous network realizes the interconnection and intercommunication among different devices of the sensor network, and solves the problem that information nodes of the heterogeneous network are difficult to perform fusion transmission by using a fusion channel. The interconnection and intercommunication effectively expands the sensing range of the Internet of things and the type of the accessed equipment, so that the sensing layer equipment and the data types are more diversified. On the other hand, different sensing node resources are internally combined, so that the complementarity among different resources can be enhanced, the advantages of various resources are fully exerted, and the comprehensive sensing capability of a sensing layer is improved; the multi-sensor network deployment mode can provide various different services for users, and better meets the requirements of various industries in the future.

The deployment observation of multiple network nodes and the perception of the heterogeneity of network resources both provide great challenges for the management and control of resources of the internet of things. The network resource management and control is the work of dynamically changing and adjusting the operation state, the working mode, the communication link, the task scheduling and the like of the sensing network equipment by the application layer. The resource management and control are mainly in the aspects of routing protocols, communication link optimization, resource positioning, network coverage optimization and the like. The basic control is only limited in the sensing network, the application layer does not directly interfere with the internal characteristics of the sensing network, and the reusability of node resources and the end-to-end transparent control of the application on the resources are more realized.

Because the nodes of the internet of things are numerous, the node deployment scale and the node operation condition are generally required to be adjusted according to the application requirements. The application layer needs to capture the frequent network state change situations in time and make corresponding observation records and remote control. For remote observation application, resource management and control are performed in a manual configuration mode, time and labor are consumed, and real intelligent observation is difficult to realize. To realize intelligent observation and dynamic deployment control, an access platform is required to be capable of dynamically adapting to various node resources, packaging and abstracting resource configuration information and data information, so that the problems of various resource access, state change and the like can be efficiently dealt with in real time while observing network operation, and the cost of sensing network deployment and maintenance is reduced.

In summary, the multi-sensor internet of things has a wide application prospect, but the multi-sensor internet of things deploys management and control middleware, and there are technical difficulties to be solved urgently at the same time, including:

(1) the internet of things has fixed communication distance, transmission bandwidth and network topology under the single perception network environment, cannot simultaneously satisfy the perception of different types of data, the application environment conditions are complex and changeable, the types and data quantity of collected data are continuously increased, only a single sensing network is deployed, the current application requirements cannot be met, the management and control of the resources of the internet of things are greatly challenged by multi-network node deployment observation and the heterogeneity of the sensing network resources, an application layer cannot dynamically change and adjust the running state, the working mode, the communication link, task scheduling and the like of sensing network equipment, the resource management and control is lack of routing protocols, communication link optimization, resource positioning and network coverage optimization and is only limited in the sensing network, the internal characteristics of the sensing network cannot be directly interfered by the application layer, and the reusability of node resources and the end-to-end transparent management and control of the application to the resources cannot be realized.

(2) Because the nodes of the internet of things are numerous, the node deployment scale and the node operation condition are generally required to be adjusted according to the application requirements. The current application layer can not timely capture the frequent network state change conditions and make corresponding observation records and remote control, the prior art uses a manual configuration mode for remote observation to control resources, which is time-consuming and labor-consuming, is difficult to realize real intelligent observation, intelligent observation and dynamic deployment control can not be realized, an access platform can dynamically adapt to various node resources, resource configuration information and data information are lacked for encapsulation and abstraction, the problems of various resource access and state change and the like can not be effectively solved in real time when the network operation can not be observed, a fault location and node operation state monitoring mechanism is lacked, the node working fault and working mode abnormity existing in the sensing network can not be timely and effectively discovered, and the sensing network deployment and maintenance cost is high.

(3) The existing network management and control under various wireless sensor network monitoring environments is complex, nodes are intermittently invalid, sensor networks under different application scenes lack effective rapid dynamic deployment means, the node resource management and control of the internet of things based on multi-sensor network fusion observation lacks improvement in the aspects of node resource identification, resource registration updating, node working mode management and control, node running state feedback and the like, does not meet the requirements of an application end for mastering network state change in real time and dynamically and rapidly deploying an observation network, middleware does not support simultaneous access of node resources of four typical heterogeneous wireless communication networks, namely ZigBee, Bluetooth, WiFi and 6Lowpan, resource management and control are not integrated, a message agent function based on an MQTT protocol is not realized, the advantage complementation of common wireless sensing networks cannot be realized, the reusability and expandability of the sensing networks are poor, an application layer cannot realize intelligentization of end-to-end and end, The node resources are managed and controlled transparently, pre-analysis and calculation of all perception data in the edge device of the Internet of things are lacked, observation data with higher value cannot be transmitted to an application end, data transmission cost and energy consumption of the device in the data transmission process are high, and delay of application decision is long.

(4) The current middleware of the Internet of things is also concentrated on the aspects of perception and interconnection of a bottom layer, and has great difference from the aspect of realizing high flexibility, reusability and high reliability of the application of the Internet of things; the application complexity of the internet of things is low, the scale is in a primary stage, the application of the internet of things supporting large-scale equipment access still has the problems of heterogeneous physical equipment fusion, service quality control, complex event processing, safety, access control and the like, the workload of operation and control is large, the functions of starting up and starting down the underlying network and the application layer, task demand programming interface fusion and the like cannot be realized, the terminal resources of the internet of things are limited, the energy is limited, the node heterogeneity and the like are high, the middleware scheme has no light weight in performance and cannot meet the control requirements of various heterogeneous node resources, the middleware has low hardware performance, cannot bear a large number of forwarding tasks of data, lacks a module for incorporating edge analysis processing of the data into the middleware function, and cannot play the role of starting up and starting down the middleware in an internet of things architecture.

Disclosure of Invention

According to the method, the problems that network management and control are complex, nodes are intermittently invalid under the condition of multiple wireless sensor network monitoring environments, sensor networks lack effective rapid dynamic deployment means under different application scenes and the like are combined, a multi-sensor network dynamic deployment and control middleware scheme and an edge data analysis method based on environment proximity calculation are successfully researched and developed, and the management and control of the node resources of the Internet of things based on the fusion observation of the multi-sensor network are centralized on the aspects of node resource identification, resource registration updating, node working mode management and control, node running state feedback and the like. The control levels all meet the requirements of an application end for mastering network state change in real time and dynamically and quickly deploying an observation network; the resource management and control and the message agent function based on the MQTT protocol are integrated, so that the advantage complementation of a common wireless sensing network is realized, the manual deployment complexity of the sensing network is reduced, the reusability and the expandability of the sensing network are improved, and the end-to-end intelligentization and transparentization of the application layer are realized to manage and control the node resources; all perception data are pre-analyzed and calculated in the middleware Internet of things edge device, observation data with higher utilization value are transmitted to an application end, data transmission cost and energy consumption of the device in the data transmission process are reduced, delay of application decision is reduced, communication distance, transmission bandwidth and network topology expansibility are strong, and perception requirements of different types of data are met.

In order to realize the technical characteristics and effects, the technical scheme adopted by the application is as follows:

the multi-sensor Internet of things resources are self-adaptive deployment control middleware and divided into a part A, namely a wireless sensor node resource virtualization control model, and a part B, namely virtualized sensor node dynamic deployment control;

wherein part A comprises: a1-node digital identity fusion mark; a2-node resource abstract description, comprising: the method comprises the following steps that an Internet of things sensor node description model and a virtual sensor abstract pool are obtained; a 3-message broker based end-to-end regulation, comprising: end-to-end resource management and control, message dynamic publishing subscription model and resource management and control model based on gravity center;

wherein part B includes: B1-Internet of things node dynamic access method, comprising: node virtualization process, dynamic data access and processing; b2-sensor node deployment management and control, comprising: an Internet of things node discovery mechanism and an Internet of things node deletion mechanism; b3-sense node operational status monitoring, comprising: link fault type division and node running state monitoring process; b4-node working mode self-adaptive adjustment, comprising: classifying sensor working modes and adjusting node working modes; b5-edge management and control analysis of data;

firstly, designing four different communication networks simultaneously accessed to ZigBee, 6Lowpan, Bluetooth and WiFi, carrying out fusion digital identity identification on each sensor node resource under the four networks, and carrying out fusion access and end-to-end management and control on an application end;

secondly, performing software virtualization description on each node and a node data processing process, converting each node of a sensing network into a virtual sensor abstract pool, shielding various isomerism of sensing layer resources, immediately accessing and adapting a system when the sensing layer resources are added or changed, fusing complex sensor original data with different structures into metadata with fixed attributes, enhancing the semanteme of the data, and simplifying the storage and transmission modes of the data;

thirdly, on the basis of an Internet of things MQTT protocol, a middleware is used as a communication agent between an application end and nodes, an end-to-end control flow is designed when the nodes of the application end are found and deleted, the running state of the nodes is monitored in a sensing mode, and the nodes are regulated and controlled in a working mode, data preprocessing and analysis allocation are added into the middleware, and transparent control of the nodes by an application layer is realized in cooperation with a communication agent mechanism;

fourthly, based on the data processing environment close computing of the Internet of things, data filtering computing is added into the middleware, and therefore the network edge equipment can execute preliminary data computing and analyzing tasks.

Preferably, the resource virtualization management and control model of the wireless sensor node comprises: the physical node resources are virtualized into virtual sensors convenient to process through software description in an abstract packaging mode, wherein an abstract model is divided into a fusion resource description model and an end-to-end management and control model, the fusion resource description model carries out centralized agreement on node attributes, data processing and storage, attributes with unique identification physical nodes and sensors carried by the nodes correspond to a data sensing process, and abstraction of hardware nodes is realized; the latter is the description of different levels of management and control modes of the application end, so that the application end integrates end-to-end management and control on different nodes, and in the two models, the middleware is used as a communication agent to be responsible for forwarding various data and management and control messages.

Preferably, the node digital identity fusion identifier is: the node digital identity + NetSort resource identification method is provided, wherein the node digital identity is the MAC address or IP address of the node when the equipment leaves a factory, and NetSort is the communication network type of a certain node and marks the network communication type of the node;

the method comprises the steps that the device address characteristics of each wireless network are integrated, the middleware is used for determining and accessing node resources under a certain network in a fusion mode, and the middleware identifies nodes of a sensing layer by resource digital identity information;

NetSort is described by one byte and is determined by different data access channels of the middleware, the node digital identity is an MAC address or an IP address of a node under each network, the node establishes communication connection with the middleware and then sends Res digital identity to the middleware, the middleware records the Res digital identity, and a digital identity description model is fused to ensure that the middleware can distinguish a communication network where the node is located and the node is a specific device under the network, and an application end masters the Res digital identity of each node and then sends control information of the Res digital identity to a designated node end to end.

Preferably, the node resource abstract description is:

1. internet of things sensor node description model

The virtual sensor description model facing the multi-sensor improved XML, wherein the middleware encapsulates each attribute and data processing process of an entity sensor into a virtual sensor in an XML file form, and the virtual sensor comprises necessary sensor node deployment and use information, and comprises the following steps: 1) a node measurement type; 2) sensor metadata information; 3) converting and processing interface information by data flow; 4) a filtering interface of the data flow and an SQL query storage flow; 5) analyzing interface information of data edge; one or the same type of sensor resource is replaced by one virtual sensor, the middleware realizes sensor virtualization and dynamically processes the resource access process.

Aiming at a middleware resource virtualization task and an end-to-end management and control task of an application end, inducing the content of metadata information of a sensor, wherein the metadata information comprises node resource digital identity, a unit of measurement data, a sensor observation type, a measurement range, a sampling period, a node running state and necessary time attributes;

for a sensor with a single function or a plurality of sensors needing cooperative work, a virtual sensor is used for carrying out fusion description on a plurality of sensors, namely information of each sensor is concentrated in an XML file for description, and heterogeneous sensor data of different sources are subjected to fusion processing and storage through a data stream conversion and data storage process defined in the virtual sensor;

2. virtual sensor abstraction pool

The method comprises the following steps that a certain sensor node or a sensor node of the same type uses a virtual sensor description file to carry out software abstract description, when a sensing network is accessed to nodes of various types, a middleware uses different virtual sensor description files to describe each type of node, the combination of the virtual sensor description files is used as a virtual sensor resource pool, the virtual sensor resource pool is used as a logic view of the whole sensing network, the virtual sensor description files corresponding to the sensing nodes are written into the middleware in advance, and each entity node resource is accessed to the middleware and then matched with a corresponding virtual sensor in an internal virtual sensor abstract pool to carry out data processing analysis, storage and forwarding;

the existing description model in the virtual sensor resource pool can be directly used by other middleware, so that resource sharing among the middleware is realized, and repeated programming when different applications are independently used for deploying and controlling the same type of resources in the utilization process of the sensor network resources is avoided.

Preferably, based on end-to-end regulation of message brokers:

1. end-to-end resource management and control

Resource management and control are divided into four aspects: the method comprises the following steps that an application end manages and controls node resource deployment, node working mode management and control, node running state monitoring and edge data edge analysis and management based on environment proximity calculation, wherein the management and control of the four aspects cover application layer management and control tasks, and the node resource deployment management and control is registration and deletion of node resources; the working mode management and control is that the application end manages and controls the node working mode according to the management and control instruction of the sensor; the monitoring of the node running state is to monitor whether the bidirectional communication between the node and the application layer in the sensing network fails; the data edge analysis means that the sensing data is directly subjected to basic analysis and calculation in a middleware, the whole management and control is carried out by an application end, the middleware is used as an auxiliary platform, the node working mode management and control, the node running state monitoring and the resource deployment management and control belong to the direct management and control of the application end, and the data edge analysis task is executed by the middleware;

the management and control of end-to-end resources are realized by bidirectional communication of a sensing layer and an application layer, a reliable communication mechanism is established between the application end and node resources, so that the application end can fuse and control any node resource and a communication network, a management and control channel is not required to be independently opened for specific resources, a message publishing and subscribing mechanism of an internet of things MQTT communication protocol is used as a prototype, and a heterogeneous resource fusion and control model based on gravity center improvement is adopted;

2. dynamic publishing and subscribing model for message

The method comprises the steps that nodes and application ends in a sensing network are used as message publishers or subscribers, a middleware is used as a message agent, sensor nodes continuously publish collected data with identification to the middleware, the collected data are received by subscription of a supply end, when the application ends need to control the nodes, control information is published to the middleware and is forwarded to a designated node by the middleware, the whole message communication is conducted according to different gravity centers of different definitions of all message communication purposes, a gathering connection mode is adopted in combination with a multi-sensor network monitoring scene, all the nodes are connected with the application ends through the middleware, the middleware is convenient to process, and a fused message communication model and a two-way message transmission mechanism are adopted to ensure that the intermediate platform can play a role in controlling service agents between the application ends and the sensing nodes on the basis of resource virtualization description;

3. resource management and control model based on gravity center

Respectively designing a gravity center name in a control message according to four aspects of control of an application end node, wherein the gravity center is a character string of UTF-8, classifying and filtering the message according to the gravity center of a subject by a platform for realizing message agent service, grading the composition of the gravity center, sending a publish message of the gravity center name by a sensor node to transmit data to an intermediate message agent, acquiring a data type contained in the message according to the gravity center name, and simultaneously subscribing a plurality of gravity center messages by using a gravity center wildcard in MQTT when the application end subscribes the message;

based on MQTT protocol, an application end and a node message communication mechanism based on gravity center is designed, node data sending and application end management and control information two types of Publish messages and Subcribe messages are calibrated by using the set theme gravity center, and when the node needs to upload sensing data and the application end needs to issue a management and control command, messages with specified gravity center names are sent for the other party to subscribe. According to different messages to be transmitted, message topics are classified:

(1) node Publish message and application Subcribe message subject calibration using sensing data as content

When the nodes work normally, sensing data are continuously sent to the middleware, at the moment, the internal part of each node periodically and circularly issues Publish messages, the messages contain the sensing data acquired by the nodes, the subject names of the publishing messages containing the specifically acquired data are determined by the measurement types of the sensors specifically connected with the nodes and the communication network types to which the nodes belong, and the subjects of the data acquisition messages are in the format of 'observer/resource digital identity/virtual sensor name';

(2) end-to-end management and control message subject calibration

Aiming at the aspects of node resource deployment, running state monitoring and working mode adjustment in the control interaction process of the application end to the perception network, the topic of each publishing or subscribing message marks specific node resource digital identity and message intention, and after the message topic is defined, the middleware performs message topic matching to complete the forwarding of each control message.

Preferably, the mechanism for discovering the tie node: after a set of nodes are designed in the form of theme messages and added into a perception network, an interactive process of authorized access is carried out when an application layer operates:

the first step is as follows: a node to be added is firstly connected with a middleware message agent platform and sends a Connect message, the message in MQTT needs to indicate the digital identity of a sender, and Res digital identity identification is carried out by a node digital identity description model, the connection is kept all the time unless the node to be added is powered off or a Disconnect message is actively sent to the middleware;

the second step is that: each node issues a registration message to the middleware after the connection is established, and the name of the center of gravity is Resg/Res digital identity;

the third step: the application end runs a process of subscribing messages, the process continuously sends a Subscribe message containing the gravity center of the wildcard, the name of the gravity center is Resg/#, the application end can be ensured to Subscribe the registration message issued by any node, and at the moment, the middleware pushes the registration messages of two different sources to the application end after the application end sends the subscription message;

the fourth step: after capturing that a node is added into a network, an application end verifies the Client digital identity, namely the Res digital identity in the message, if the digital identity is illegal or the application does not need to add the node, directly replying an unsubcribe message to a middleware, and directly ignoring the node by the middleware; if the digital identity is legal and the node meets the requirement, the application end immediately issues an authorization message and receives the registration request of the node, the message subject is a ResgAck/Res digital identity, and the Res digital identity is the digital identity of a specific node allowed to be accessed.

The fifth step: the node issuing the registration message subscribes the authorization message after a period of time, the subject is the ResgAck/Res digital identity, and only the really authorized node receives the authorization message issued by the application end;

and a sixth step: and after the node is authorized to access, starting to push own sensing data.

The above interaction process has two preconditions: firstly, before a node to be added works normally, a registration message needs to be issued, and the subject name is expressed as a Resg/node Res digital identity; secondly, the application terminal needs to subscribe the registration message and obtains Res digital identity of the node to be registered, and the topic of the subscription message comprises a wildcard; the application end and the middleware record Res digital identities of all nodes authorized to be added, and all Publish messages and Subcribe messages do not contain specific load data (payload) before the nodes receive the authorization messages; after the authorization is successful, the nodes push own perception data to the middleware.

Preferably, the mechanism for deleting the internet of things comprises: the application end actively initiates a process of deleting indication according to the change of the perception task or the abnormal perception of the node, and the process comprises three steps:

step 1: the application end firstly determines the node digital identity of the node to be deleted and sends Publish information with the gravity center of Delete/Res digital identity to the middleware;

step 2: a node which normally works initiates subscription to the middleware at regular time to determine whether a deletion instruction from an upper layer exists or not, wherein the center of gravity of a subscription message is a Delettc/Res digital identity, and the Res digital identity is a digital identity mark of the node;

and 3, step 3: after the middleware successfully publishes the subscription theme, a certain node can successfully subscribe a deletion instruction, actively initiates Disconnect connection to the middleware to Disconnect message communication with the middleware.

When the node is deleted, the deletion indication message of the application end needs to be published first, and the node needs to subscribe the Subcribe message with consistent theme when successfully receiving the deletion indication, the message agent does not process when subscribing the message with the nonexistent theme name in the MQTT protocol, and once the subscribed message is published, the message is immediately forwarded to the subscription node, so that the node needs to actively and periodically subscribe the message with the digital identity with the Delete/Res center of gravity to the middleware when the node works normally, and the disconnected node does not perform the data sensing task any more.

Preferably, the node operation state monitoring process: the application layer judges the specific node running state according to the data receiving condition and records the fault state in the equipment state table, the application end firstly detects whether the sensing data is received, when no data is received, the application end can send a Connect message in an MQTT protocol to determine whether the middleware can be connected, if the CONNACK packet of the connection response is not received, the middleware connection is interrupted, the middleware can be recovered after the reconnection, and the connection response packet is not received, namely the middleware has faults;

when the application layer receives data, the middleware is normal in operation and can perform data processing and message agent functions, and when the time interval of the received data is not a fixed acquisition period, the node sensing task can be determined not to be normally completed; if other node data are received and the node does not transmit data within a time period, determining that the node cannot communicate with the middleware and recording as a Disconnected state of the node;

and by adopting the advice message, when the node and the middleware are disconnected suddenly, the application end finds the phenomenon that the node is disconnected suddenly in time.

After a node in a sensing network initiates connection to the middleware for the first time, a testament message is immediately sent to the middleware, the gravity center name of the message is 'Disconnected/Res digital identity', and an application end subscribes the message with the gravity center name of 'Disconnected/#' to the middleware and is used for receiving the testament messages of all nodes. The information of the will order is sent to the middleware only once, the information will be stored by the middleware until the middleware finds that the node is disconnected, the information will be immediately transferred to the application end subscribing the information, and the application end determines that the sensing node whose node digital identity is Res digital identity exits from the connection due to abnormality.

Preferably, the node working mode is adaptively adjusted:

1. sensor operating mode classification

The type 1 is related to sensor setting, and comprises node data output type setting and function mode control, wherein the control depends on a control protocol specified in the sensor, and an application end sends a specific control message to a remote node according to the control protocol to change the working mode of the remote node;

the class 2 is related to task requirements, including data acquisition frequency, sensor attitude, restart or dormancy time, and the management and control is related to decision of an application end and is suitable for all types of sensor nodes.

2. Node working mode adjusting process

The working mode adjustment is a process that an application layer actively and perceptibly initiates end-to-end communication with a network node, the transparent control is realized in that the application layer only needs to acquire the Res digital identity of the node to be controlled, the control message is forwarded to a specific node under a specific network and is replaced by a middleware, and the working mode adjustment of the node needs to ensure that the operated node is in a communicable state:

the method comprises the following steps: the application end firstly inquires out Res digital identity of a node to be adjusted from a node information table, and sends a Publish message to the middleware, wherein the subject center of gravity of the Publish message is Adjust/Res digital identity, a Payload field of the Publish message is node control information, the information is determined according to specific adjustment content, and when a control protocol of a sensor manufacturer is required to be used for control, a control protocol packet is directly filled in the Payload field;

step two: each running node regularly subscribes a data packet with the gravity center of 'Adjust/Res digital identity' to the middleware, the Res digital identity is the identifier of the node, once an application end issues a working mode adjusting message, Payload content of the message is transmitted to the node, and the node is adjusted to a new working mode according to the content;

step three: in order to ensure that the application end can determine that the working mode adjustment message is successfully received by the node, the following settings are made inside the node: the node regularly subscribes the working mode adjusting message, and immediately issues a message with the topic of 'changed/Res digital identity' for the application terminal to subscribe once the subscription is successful, and the message informs the application terminal that the state of the node is successfully changed;

step four: the application end receives all the node state Changed messages with the adjusted working mode by using a universal center of gravity, namely subscription messages of 'Changed/#';

if the application does not receive the information of the changed state sent by the sensing node, the process is repeated, and the monitoring process of the running state of the sensing node is carried out to determine whether the node or the middleware has a fault.

Preferably, the edge governance resolution of the data: based on the middleware, environment proximity calculation is adopted, and basic data management and control tasks are finished by the middleware, so that the management and control pressure and the data processing tasks of an application end are reduced, and the decision delay is reduced;

after the middleware receives the original data of each sensing node, the original data is analyzed into uniform metadata, namely actual measurement values, according to a data analysis interface in the virtual sensor description file, and before the measurement values are stored and forwarded, the following processing is carried out:

step 1: invalid data are filtered, invalid data can appear after the original node data are analyzed, and the data are not stored in a local database of the middleware and are not forwarded to an application end, so that the data forwarding amount is reduced, and the bandwidth occupancy rate is reduced;

and 2, step: the forwarding frequency is controlled, and unnecessary data uploading tasks are reduced by controlling the data forwarding frequency by the middleware;

and step 3: the method comprises the steps that a timestamp is marked, a clock module is not arranged in a sensor node, a node data stream can be marked with the timestamp only after being received by a middleware, and the node data is added with a time attribute and then stored and forwarded, so that more complex data analysis can be performed subsequently;

and 4, step 4: and (3) calculating in real time, wherein the middleware locally stores a specified number of data records, the records are concentrated in a certain time period, and the mean value and the extreme value of the locally stored data in the time period are calculated through database query processing for anomaly analysis.

Compared with the prior art, the innovation points and advantages of the application are as follows:

(1) according to the method, the problems that network management and control are complex, nodes are intermittently invalid under the condition of multiple wireless sensor network monitoring environments, sensor networks lack effective rapid dynamic deployment means under different application scenes and the like are combined, a multi-sensor network dynamic deployment and control middleware scheme and an edge data analysis method based on environment proximity calculation are successfully researched and developed, and the management and control of the node resources of the Internet of things based on the fusion observation of the multi-sensor network are centralized on the aspects of node resource identification, resource registration updating, node working mode management and control, node running state feedback and the like. The control levels all meet the requirements of an application end for mastering network state change in real time and dynamically and quickly deploying an observation network; the middleware supports simultaneous access of four typical heterogeneous wireless communication network ZigBee, Bluetooth, WiFi and 6Lowpan node resources, and integrates resource management and control and a message agent function based on an MQTT protocol, so that not only is the advantage complementation of a common wireless sensing network realized, the artificial deployment complexity of the sensing network is reduced, the reusability and expandability of the sensing network are improved, but also the node resources are managed and controlled by the end-to-end intelligentization and transparence of an application layer; in addition, all perception data are pre-analyzed and calculated in the middleware Internet of things edge device, observation data with higher utilization value are transmitted to an application end, data transmission cost and energy consumption of the device in the data transmission process are reduced, delay of application decision can be reduced, communication distance, transmission bandwidth and network topology expansibility are strong, and perception requirements of different types of data are met.

(2) The middleware of the Internet of things enhances the expandability and reusability of a sensing network, reduces the operation and control workload, realizes the functions of starting and stopping the underlying network and the application layer, integrating task demand programming interfaces and the like, has corresponding platform support for strong services, and meets the requirements of the middleware scheme on the characteristics of limited terminal resources, limited energy, high node heterogeneity and the like of the Internet of things on light weight in performance and can meet the control requirements of various heterogeneous node resources; meanwhile, the middleware has high hardware performance, can bear a large amount of data forwarding tasks, and brings edge analysis processing of the data into a middleware functional module, so that the bright spot can better play a role of the middleware in the Internet of things architecture, and high flexibility, reusability and reliability of the application of the Internet of things are realized; the application of the Internet of things supporting large-scale equipment access also has the advantages of greatly improving the aspects of heterogeneous physical equipment fusion, service quality control, complex event processing, safety, access control and the like.

(3) The method is designed and constructed for self-adaptive deployment and control middleware of multi-sensor Internet of things resources, and dynamic decision of data is carried out based on environmental proximity calculation; classifying the current data acquisition platform based on the networking communication technology and the difference of node equipment of a sensing layer, and designing middleware to realize the fusion access deployment of four common wireless communication network nodes; through virtualization of sensor node attributes and processing in the sensing network, a virtual sensor abstraction pool is formed inside the middleware platform, hardware abstraction of the sensing network is completed, implementation details of heterogeneous sensing nodes and communication protocols are shielded, and deployment of the nodes in operation is achieved; based on an Internet of things MQTT protocol, a real-time dynamic management and control message publishing and subscribing process of an application terminal in the aspects of node resource discovery and deletion, node running state monitoring, node working modes and the like is designed and realized, end-to-end interaction between sensor nodes and the application terminal is realized, and an application layer can transparently manage and control different nodes; based on the data processing environment of the internet of things is close to the calculation, data filtering and calculation are added in the middleware, so that network edge equipment can execute preliminary data calculation and analysis tasks, interconnection and intercommunication among different equipment of a sensor network are realized, the problem that heterogeneous network information nodes are difficult to perform fusion transmission through fusion channels is solved, the sensing range of the internet of things and the type of the accessed equipment are expanded, sensing layer equipment and data types are more diversified, different sensing node resources are internally combined, complementarity among different resources is enhanced, the advantages of various resources are fully exerted, the comprehensive sensing capability of a sensing layer is improved, various different services are provided for users, and the requirements of various industries in the future are better met.

(4) The method adopts an abstract packaging mode to virtualize physical node resources into virtual sensors convenient to process through software description, integrates a resource description model to carry out centralized agreement aiming at node attributes, data processing and storage, and corresponds the attributes and data sensing processes of the physical nodes with unique identification and the sensors carried by the nodes to realize the abstraction of hardware nodes; the end-to-end control model is the description of different levels of control modes of the application end, so that the application end fuses the end-to-end control of different nodes; the method comprises the steps of shielding various isomerism of a sensing network through a virtual sensor description model, fusing and describing access and data processing conversion and storage processes of different types of nodes, and performing fusion identification on digital identities of various network nodes before model abstraction so that middleware and upper-layer application can access the nodes by fused Res digital identities; the end-to-end communication adopts an application layer protocol in the Internet of things based on the MQTT-SN and MQTT protocols of a publishing and subscribing mechanism to carry out targeted calibration on the gravity centers of messages of different communication purposes, so that node messages from different networks and control messages of an application end can be matched through the message theme of a middle platform, and the end-to-end remote control of sensing nodes by the application end is realized. The method has the advantages that the intermediate layer shields the heterogeneity of the data source of the sensing layer, the system development cost brought by resource updating is reduced, the application layer can effectively deal with the problems of various resource accesses, state changes and the like in real time while observing the network operation by transparent control node equipment, and the deployment and maintenance cost of the sensing network is reduced.

Drawings

Fig. 1 is a schematic diagram of a node digital identity description model designed in the present application.

FIG. 2 is a flow chart of the selection of trajectory predictors and initial crowd-sourcing fitting centers.

FIG. 3 is a logical diagram of virtual sensor to actual sensor node correspondence for a virtual sensor abstraction pool.

FIG. 4 is a diagram of a publish and subscribe message topic for a node that perceives data as content.

FIG. 5 is a resource end-to-end governing message topic model based on a center of gravity.

Fig. 6 is a schematic flow chart of node abstraction of the internet of things into a virtual sensor.

Fig. 7 is a schematic diagram of an application node operation state monitoring record.

FIG. 8 is a schematic diagram of a node operational status monitoring process.

Fig. 9 is a schematic diagram of an interaction process between an application and a node when a node operation mode is adjusted.

Fig. 10 is a schematic diagram of an edge management parsing process of data.

Detailed description of the invention

Specific embodiments of the self-adaptive deployment management and control middleware of the multi-sensor internet of things resource are described in detail below with reference to the accompanying drawings, so that those skilled in the art can better understand and implement the application. Those skilled in the art should appreciate that they can readily use the present disclosure as a basis for designing or modifying other structures for carrying out the same purposes of the present disclosure and that such modifications are intended to be included within the scope of the present disclosure.

The wireless sensor network is used as an important foundation and a component of the Internet of things, is widely deployed in various applications such as environment monitoring, smart cities, modern agriculture and smart homes, and is responsible for acquisition and transmission of the sea sensing data in various applications. Because different sensor resource data protocols and physical interfaces are different, wireless networking communication technologies are rich and various, most applications integrate the differences according to respective mechanisms and application characteristics, and an integrated internet of things system which is suitable for various node resource attributes and can be flexibly deployed and transparently controlled is not established. The whole internet of things sensing system has to be redeployed after the application scene and the requirement are changed, so that the difficulty and the cost of the internet of things deployment in a large-range sensing scene are greatly increased, and the flexibility and the efficiency of the whole internet of things system are difficult to embody. In addition, under a cloud computing architecture, a large amount of transmission bandwidth is occupied by a mode of transmitting mass sensing data to a cloud end, timeliness of data processing and decision making is reduced, and the trend that data and equipment of the internet of things grow rapidly is difficult to deal with.

In order to solve the defects existing in the application of the Internet of things, the application provides an Internet of things middleware.

(1) And simultaneously accessing four different communication networks of ZigBee, 6Lowpan, Bluetooth and WiFi, fusing digital identity identifiers for each sensor node resource under the four networks, and performing fusion access and end-to-end management and control on an application terminal.

(2) The method comprises the steps of performing software virtualization description on each node and a node data processing process, converting each node of a sensing network into a virtual sensor abstract pool, shielding various isomerism of sensing layer resources, solving the problem that a system cannot be immediately accessed and adapted when the sensing layer resources are added or changed, fusing complex sensor original data with different structures into metadata with fixed attributes, enhancing the semanteme of data, and simplifying the storage and transmission modes of the data.

(3) Based on an application layer protocol MQTT of the Internet of things, a middleware is used as a communication agent between an application end and nodes, an end-to-end control flow of the application end in node discovery and deletion, node operation state monitoring sensing and node working mode adjustment and control is designed, data preprocessing and analysis allocation are added into the middleware, and transparent control of the application layer on the nodes is realized in cooperation with a communication agent mechanism.

First, a wireless sensor node resource virtualization control model

The design intermediate layer shields the heterogeneity of the data source of the sensing layer, reduces the system development cost brought by resource updating, and enables the application layer to transparently control node equipment. The method comprises the steps that physical node resources are virtualized into virtual sensors convenient to process through software description in an abstract packaging mode, wherein an abstract model is divided into a fusion resource description model and an end-to-end management and control model, the abstract model carries out centralized agreement on node attributes, data processing and storage, attributes and data sensing processes of the physical nodes and the sensors carried by the nodes with unique identification are corresponding, and abstraction of hardware nodes is achieved; the latter is the description of different levels of management and control modes of the application end, so that the application end integrates end-to-end management and control on different nodes, and in the two models, the middleware is used as a communication agent to be responsible for forwarding various data and management and control messages.

The method comprises the steps of shielding various isomerism of a sensing network through a virtual sensor description model, fusing and describing access and data processing conversion and storage processes of different types of nodes, and fusing and identifying the digital identities of all network nodes before model abstraction so that middleware and upper-layer application can access all nodes with fused Res digital identities.

The end-to-end communication adopts an MQTT-SN and MQTT protocol based on a publish and subscribe mechanism in an application layer protocol in the Internet of things, on the basis of the MQTT protocol, the gravity centers of messages of different communication purposes are calibrated in a targeted manner, so that node messages from different networks and control messages of an application end can be matched through the message theme of a middle platform, and the end-to-end remote control of sensing nodes by the application end is realized.

Node (I) digital identity fusion identifier

When multiple wireless sensor networks are simultaneously distributed in the sensing layer, each node has a unique device identifier, such as a device MAC address in a ZigBee network, an IPv6 address of a node in 6Lowpan, and the like. In the node identification method in the prior art, when sensing data coding, equipment digital identity numbers are manually added, so that the complexity of data processing is increased, and real resource identification and distinguishing cannot be realized. The MAC address of the node is directly used to access the node, which does not guarantee that the management and control information of the application end can be transmitted to the designated node of the corresponding sensing network. Therefore, the application provides a resource identification method of node digital identity + NetSort, wherein the node digital identity is an MAC (media access control) address or an IP (Internet protocol) address of a node (for ZigBee equipment and Bluetooth equipment) when equipment leaves a factory, and NetSort is a communication network type to which a certain node belongs and marks the network communication type to which the node belongs.

Compared with manual node number marking, the node identification method can better ensure that the application end accurately and reliably positions the node resources under different types of perception networks and carries out end-to-end communication.

In order to determine and access node resources under a certain network in a fusion mode by integrating the equipment address characteristics of each wireless network, the node digital identity description model shown in figure 1 is designed in the application, and each node of a sensing layer is identified by the middleware through resource digital identity information.

NetSort is described by one byte and determined by different data access channels of the middleware, the node digital identity is an MAC address or IP address of a node under each network, the node sends Res digital identity to the middleware after establishing communication connection with the middleware, the middleware records the Res digital identity, a digital identity description model is fused to ensure that the middleware can distinguish a communication network where the node is located and the node is specific equipment under the network, and an application end masters the Res digital identity of each node and then sends control information of the node to a specified node end to end.

Aiming at the distinguishing problem that the middleware accesses a plurality of networks of the same type, the physical communication interface of the middleware judges that if two Zigbee networks access the raspberry group middleware, each ZigBee network data can be transmitted to the interior of the raspberry group through different serial ports of the raspberry group, and the interior of the raspberry group distinguishes different serial port equipment numbers and is used for distinguishing nodes under different Zigbee networks.

(II) abstract description of node resource

The prior art includes IEEE 1451 protocol cluster and a sensor modeling language SensorML in SWE, the former focuses on standardization of sensor interfaces and tends to design an underlying layer, and the SensorML is a sensor description model formulated based on XML, but does not specify fields of sensor address identification, and locates resources only through other geographic attributes, which is not suitable in a multi-sensor network deployment environment. In addition, the realization of the description model makes the description content vary greatly due to the subjective understanding of different modelers.

1. Internet of things sensor node description model

A virtual sensor description model for multi-sensor improved XML, wherein a middleware encapsulates each attribute and data processing process of an entity sensor into a virtual sensor in an XML file form, and the virtual sensor comprises necessary sensor node deployment and use information, and comprises the following steps: 1) a node measurement type; 2) sensor metadata information; 3) converting the data stream and processing the interface information; 4) a filtering interface of the data flow and an SQL query storage flow; 5) analyzing interface information at data edges; one or the same type of sensor resource is replaced by one virtual sensor, the middleware realizes sensor virtualization and dynamically processes the resource access process.

For the middleware resource virtualization task and the end-to-end management and control task of the application end, the metadata information content of the sensor is summarized as fig. 2, and the metadata information includes node resource digital identity, unit of measurement data, sensor observation type, measurement range, sampling period, node running state and necessary time attribute.

For a sensor with a single function or a plurality of sensors which need to work cooperatively (for example, a plurality of sensors need to perform position identification and need to work together with a GPS sensor), a virtual sensor is used for performing fusion description of a plurality of sensors, that is, information of each sensor is described in an XML file, and heterogeneous sensor data from different sources are processed and stored in a fusion manner through a data stream conversion and data storage process defined in the virtual sensor, so that data processing and control in the later stage are facilitated.

2. Virtual sensor abstraction pool

One or the same type of sensor node uses the virtual sensor description file to perform abstract description on software, when the sensing network is accessed to a plurality of different types of nodes, the middleware needs to use different virtual sensor description files to describe each type of node, the combination of the virtual sensor description files is used as a virtual sensor resource pool, the virtual sensor resource pool is regarded as a logic view of the whole sensing network, and fig. 3 shows the corresponding logic of the virtual sensor and the actual sensor node. Virtual sensor description files corresponding to the sensing nodes are written in the middleware in advance, and after each entity node resource is accessed into the middleware, the corresponding virtual sensor in the internal virtual sensor abstract pool is matched for data processing analysis, storage and forwarding.

(III) message broker-based end-to-end policing

1. End-to-end resource management and control

This application combines the characteristics of multisensor network monitoring application, divide into four aspects with resource management and control: the method comprises the following steps that an application end manages and controls node resource deployment, node working mode management and control, node running state monitoring and edge data edge analysis and management based on environment proximity calculation, wherein the management and control of the four aspects cover application layer management and control tasks, and the node resource deployment management and control is registration and deletion of node resources; the working mode management and control is that the application end manages and controls the node working mode according to the management and control instruction of the sensor; the monitoring of the node running state is to monitor whether the bidirectional communication between the node and the application layer in the sensing network fails; the data edge analysis means that the sensing data is directly subjected to basic analysis and calculation in a middleware, the whole management and control is carried out by an application end, the middleware is used as an auxiliary platform, the node working mode management and control, the node running state monitoring and the resource deployment management and control belong to the direct management and control of the application end, and the data edge analysis task is executed by the middleware.

The management and control of the end-to-end resources are realized through the bidirectional communication of the sensing layer and the application layer, a reliable communication mechanism is established between the application end and the node resources, so that the application end can fuse and control any node resource and a communication network, a management and control channel does not need to be independently opened for specific resources, a message publishing and subscribing mechanism of an internet of things MQTT communication protocol is used as a prototype, and a heterogeneous resource fusion and control model based on gravity center improvement is adopted.

2. Dynamic publishing and subscribing model for message

The nodes and the application terminals in the sensing network are used as message publishers or subscribers, the middleware is used as a message agent, the sensor nodes continuously publish the acquired data with the identification to the middleware for the subscription and receiving of the application terminals, and the application terminals can publish the control information to the middleware and transmit the control information to the designated nodes when the application terminals need to control the nodes. The whole message communication adopts a gathering connection mode by combining a multi-sensor network monitoring scene according to different gravity centers defined by different message communication purposes, all nodes are connected with an application end through a middleware implementation platform, the middleware processing is facilitated, and a control service agent between the application end and a sensing node can be exerted on the basis of resource virtualization description by adopting a converged message communication model and a two-way message transmission mechanism.

3. Resource management and control model based on gravity center

According to the four aspects of management and control of an application end node, a gravity center name in a management and control message is respectively designed, the gravity center is a character string of UTF-8, a platform for realizing message agent service can classify and filter messages according to the gravity center of a theme, the composition of the gravity center is graded, a sensor node sends publish messages of the gravity center name to transmit data to an intermediate message agent, the data type contained in the message is obtained according to the gravity center name, and when the application end subscribes the message, the gravity center wildcard in MQTT is used for subscribing the messages of a plurality of gravity centers at the same time.

In order to enable an application end to acquire a sensing data stream and intelligently manage and control nodes in real time, the application end and the nodes are designed based on an MQTT protocol, a message communication mechanism of the application end and the nodes based on gravity is designed, node data are sent, two types of Publish messages and Subcribe messages of management and control information of the application end are calibrated by using a set theme gravity, and when the nodes need to upload sensing data and the application end needs to issue a management and control command, messages with specified gravity names are sent to the other side for subscription. According to different messages to be transmitted, message topics are classified:

As shown in fig. 4, when the nodes work normally, the sensing data is continuously sent to the middleware, at this time, each node periodically and circularly issues Publish a Publish message, the message includes the sensing data acquired by the node, the subject name of the Publish message including the specifically acquired data is determined by the sensor measurement type specifically connected to the node and the communication network type to which the node belongs, and the subject of the data acquisition message is in the format of "observer/resource digital identity/virtual sensor name".

(2) End-to-end management and control message subject calibration

In the application, in the management and control interaction process of the application to the sensing network, aiming at the aspects of node resource deployment (including node registration and deletion), operation state monitoring and working mode adjustment, according to the interaction requirements of different management and control aspects, a resource end-to-end management and control model based on the gravity is listed in fig. 5.

The topic of each publishing or subscribing message marks the specific node resource digital identity and the message intention, and after the message topic is defined, the middleware performs message topic matching to complete the forwarding of each control message.

Second, dynamic deployment management and control of virtualized sensor nodes

Dynamic access method for nodes of Internet of things

1. Node virtualization process

The node resources of the sensing network are abstractly described, the sensing layer node resources are converted into a virtual sensor abstraction pool, the virtual sensor abstraction pool comprises information of access, data conversion and storage methods of all the node resources, a user writes corresponding node description files according to sensor characteristics and processing methods actually connected with each node, after the description files in the virtual sensor abstraction pool are written once, the description files are convenient to be repeatedly used when other nodes with the same function are deployed, and the repeated same configuration work when the same nodes are deployed is reduced.

The middleware carries out virtualization description on a sensor of each node accessed by a sensing network, each node is abstractly described by an XML file according to the sensor connection condition of each node, when the node is connected with only one sensor, the virtual sensor only comprises the virtual description of the sensor, when the node is connected with two or more sensors of different types, the virtual sensor comprises the virtual description of a plurality of sensors, in addition, the data analysis and processing modes of each node are different, and in order to support the rapid deployment and the data access of the node sensor, each sensor node virtualization description file comprises necessary data processing and storage methods, and comprises 1) node measurement attributes; 2) sensor metadata information; 3) data access and interface information analysis; 4) a data stream storage interface; 5) a data edge analysis and processing interface.

FIG. 6 shows the flow of the middleware to associate nodes with virtual sensor files after the nodes are added. After each node is added into the network, sensing data are transmitted to a middleware through a Publish message, the theme in the Publish message comprises the name of a virtual sensor, the middleware matches a certain virtual sensor file in an internal virtual sensor abstract pool according to the name, and if the node is added into the sensing network before, the existing virtual sensor file is directly started; and if the node is the new type of node, matching the virtual file according to the barycentric name of the Publish message sent by the new node.

After the nodes are matched with the corresponding virtualization description files, the middleware analyzes and converts heterogeneous sensor original data into measured values according to each processing interface recorded in the virtual sensor, and completes the processes of storage of the processed data and the like, wherein a data analysis function is used for converting the sensor original data acquired by the middleware into specific observation values in the process, and the function is marked in the virtual sensor description files as a Java processing class and is convenient to call; the data filtering interface and the data storage interface are marked in the virtual sensor description file, and the middleware application program automatically calls the interfaces to complete the processing process of the data when analyzing the virtual sensor description file.

2. Dynamic data access and processing

The method comprises the steps of providing access for sensing data fusion by utilizing a middleware, removing redundancy of heterogeneous data, analyzing original byte data and assisting with metadata for description, fusing and filtering the data, and transmitting the data to an application end with the least transmission quantity and bandwidth.

The dynamic processing mode of the middleware on the data stream comprises the following steps:

(1) access management and control: the middleware controls the four data access channels, determines whether to receive the sensing data of a certain network or a certain node according to requirements, and the acquired node is the original node data and exists in a byte form.

(2) Data analysis: the sensing data in the byte form converts an analysis interface in the calling virtual sensor into a readable observation value with a time stamp, the data analysis interface is compiled according to a sensor data protocol and is marked in a node virtualization description file, and the data analysis interface can be called when a group of node original data packets are received.

(3) And (3) temporarily storing data: the data storage comprises the storage of sensor data and the storage of node information, after the original data of each node is analyzed into an observed value, the observed value is temporarily stored in each table of a local database of the middleware according to a certain storage capacity, the storage capacity is continuously refreshed, only the latest storage capacity strip record is reserved, each node data is separately stored in each data table, the middleware records all accessed node metadata information, and the information is stored as a node information table.

(4) Data edge analysis: and the middleware adds a primary data anomaly detection interface and a primary data filtering interface, dynamically analyzes the observation data in the temporary storage table, finds whether the network node data is normal in time and reduces the transmission of useless data.

(5) Data fusion and forwarding: the fusion mode comprises data SQL query and node data combination, wherein the SQL query mode is based on a temporary storage table, multiple node data are extracted within a limited threshold value or the node data are counted in a certain time period, and the realization mode is that the node data to be combined are uploaded to an application end by specified data in a virtualized node description file or a specific SQL statement query mode.

(II) sensor node deployment management and control

In order to meet different monitoring requirements, an application end needs to adjust and sense the network deployment situation in real time, and discovery and deletion of node resources are involved. In order to realize dynamic network resource deployment and enable an application end to master and sense node deployment conditions in time, a node deployment control implementation mechanism based on a publish and subscribe mode and taking the application end as a main factor is designed and divided into a node discovery mechanism and a node deletion mechanism.

1. Internet of things node discovery mechanism

The sensing network structure is changed along with the change of application requirements, for example, when a monitoring area is enlarged or when the monitoring equipment of the existing network is insufficient to provide required data, a new sensing node needs to be added. And when a new node is added and the application system is ensured to maintain a normal operation state, the system is required to have enough expandability. Moreover, if the addition of the node in the sensing network is not authorized, malicious data or events will affect the security of the whole application, so a reliable node discovery mechanism is needed, which can ensure the "plug and play" of the node and can realize the safe and reliable addition of the node to the sensing network.

The MQTT does not directly realize client discovery or gravity center discovery mechanism, so that the application designs an interactive flow of authorized access when an application layer runs after a set of nodes are added into a perception network in the form of theme messages. The specific interaction process is as follows:

the second step: each node issues a registration message to the middleware after the connection establishment is completed, and the name of the center of gravity is Resg/Res digital identity;

2. Internet of things node deletion mechanism

The node deletion is a process that an application end actively initiates deletion indication according to the change of a perception task or abnormal perception of the node, and comprises three steps:

and 3, step 3: after the middleware issues the subscription theme and is successfully matched, a certain node can successfully subscribe a deletion instruction, actively initiates the disconnection of the Disconnect to the middleware to Disconnect the message communication with the middleware.

(III) monitoring of operation state of sensing node

The sensing network in the internet of things has the characteristics of large number of nodes, wide distribution range, unstable links, complex deployment environment, limited storage and calculation resources and the like, and the complex external environment and the wireless transmission quality enable the sensing network and a general network to be prone to link failure, so that the monitoring task cannot be normally carried out. Therefore, the network resources are intelligently and dynamically controlled, and meanwhile, the state of the network link is monitored in a sensing mode, so that the loss caused by the network link failure is reduced as much as possible by the application end.

1. Link failure type partitioning

The link faults are classified from two layers, and firstly, the perceived link faults are divided into the following steps from the system level: node failures and communication network failures. The former means equipment interface, processing module, signal transmission and power supply line failure, which is expressed as that the node can not collect data, has no message response, data reading and writing error, insufficient power supply and node dormancy; the communication network failure is link transmission or network connectivity failure caused by network congestion, data packet loss and limited communication distance. From the duration of the occurrence of the failure, the link failure is divided into intermittent and permanent failures. Intermittent faults are faults that are set by adjusting software, but which can be restored to normal without replacing system components. Permanent failure means that the equipment or the communication link is failed and must be repaired manually to recover to normal.

The final result of these failures is that the whole or part of the data acquisition and node management tasks cannot run normally, but it is difficult for the application end to truly lock which link or which device the failure occurs in. Therefore, a set of link state monitoring mechanism based on multi-sensor network perception is designed.

2. Monitoring process for node running state

The application layer judges a specific node operation state according to the data receiving condition and records the fault state in the equipment state table, and the specific node state monitoring record judging and state monitoring processes are respectively shown in fig. 7 and fig. 8.

When the application layer receives data, the middleware is normal in operation and can perform data processing and message agent functions, and when the time interval of the received data is not a fixed acquisition period, the node sensing task can be determined not to be normally completed; if other node data are received and no data are transmitted from the node within a time period, determining that the node cannot communicate with the middleware and recording as a Disconnected state of the node; specifically, the application determines the node operation fault and the record condition according to the data receiving condition as shown in fig. 7.

After a node in a sensing network initiates connection to the middleware for the first time, a testament message is immediately sent to the middleware, the gravity center name of the message is 'Disconnected/Res digital identity', and an application end subscribes the message with the gravity center name of 'Disconnected/#' to the middleware and is used for receiving the testament messages of all nodes. The will order message will be sent to the middleware only once, the message will be kept by the middleware until the middleware finds that the node has been disconnected (keep-alive time has arrived), and will be immediately reported to the application end subscribing the message, the application end determines the sensing node whose node digital identity is Res digital identity quits connection because of abnormality.

(IV) node working mode self-adaptive adjustment

1. Sensor operating mode classification

Besides acquiring the sensing node acquisition data, the application layer also needs to remotely control the nodes in the complex environment and change the working mode of the nodes to meet the application requirements. For example, in the ground surface monitoring application, if a disaster such as landslide and earthquake occurs in a monitoring area within a certain time period, the application end is often interested in the perception data before and after the time period, and needs to analyze sufficient monitoring data. In this case, the application may shorten the sensing node data acquisition period through remote control, and acquire the surface environment data at a higher frequency, where the control of the acquisition period is an example of adjusting the working mode of the sensor by the application.

The working mode adjustment of the sensor node is divided into two types:

the 1 st type is related to sensor settings and comprises node data output type setting and function mode control, the control depends on a control protocol specified in the sensor, and an application end sends a specific control message to a remote node according to the control protocol to change the working mode of the node;

the type 2 is related to task requirements, comprises data acquisition frequency, sensor attitude and restarting or dormancy time, is related to decision making of an application end and is suitable for all types of sensor nodes, and the management and control of the application end are convenient for reasonably planning the whole sensing network acquisition task according to specific requirements, energy consumption of a sensing layer and matching factors among nodes.

2. Node working mode adjusting process

The working mode adjustment is a process that an application layer actively and perceptively initiates end-to-end communication with network nodes, the transparent control is realized in that the application layer only needs to obtain Res digital identities of the nodes to be controlled, the control information is forwarded to specific nodes under a specific network and is replaced by a middleware, and the nodes need to be ensured to be in a communicable state by carrying out the working mode adjustment on the nodes. The specific interaction process is shown in fig. 9.

The method comprises the following steps: the application end firstly inquires out Res digital identity of a node to be adjusted from a node information table, and sends Publish information to the middleware, wherein the subject gravity center of the Publish information is Adjust/Res digital identity, a Payload field of the Publish information is node control information, the information is determined according to specific adjustment content, and when a control protocol of a sensor manufacturer is required to be used for control, a control protocol packet is directly filled in the Payload field;

step three: in order to ensure that the application end can determine that the working mode adjustment message is successfully received by the node, the following settings are made inside the node: the node subscribes the adjusting message of the working mode periodically, and immediately issues a message with the topic of 'changed/Res digital identity' for the application terminal to subscribe once the subscription is successful, and the message informs the application terminal that the state of the node is successfully changed.

Step four: the application end receives all the node state Changed messages with the adjusted working mode by the universal center of gravity, namely the subscription message of 'Changed/#'.

(V) edge management and control analysis of data

The application adopts environment proximity calculation based on the middleware, and basic data management and control tasks are completed by the middleware, so that the management and control pressure and the data processing tasks of an application end are reduced, and the decision delay is reduced.

After receiving the raw data of each sensing node, the middleware parses the raw data into uniform metadata, i.e., actual measurement values, according to a data parsing interface in the virtual sensor description file, and before storing and forwarding the measurement values, the middleware performs the following processing, as shown in fig. 10:

step 1: invalid data are filtered, and after the node original data are analyzed, invalid data can appear, and the data cannot be stored in a local database of the middleware or forwarded to an application end, so that the data forwarding amount is reduced, and the bandwidth occupancy rate is reduced;

step 2: the forwarding frequency is controlled, and unnecessary data uploading tasks are reduced by controlling the data forwarding frequency by the middleware;

Claims

1. The method is characterized by comprising the following steps that (1) multi-sensor Internet of things resources are adaptively deployed and controlled middleware, and the middleware is divided into a part A, namely a wireless sensor node resource virtualization control model, and a part B, namely virtualized sensor node dynamic deployment control;

secondly, performing software virtualization description on each node and the node data processing process, converting each node of a sensing network into a virtual sensor abstract pool, shielding various isomerism of sensing layer resources, immediately accessing and adapting a system when the sensing layer resources are added or changed, simultaneously fusing original sensor data with complex and different structures into metadata with fixed attributes, enhancing the semanteme of data, and simplifying the storage and transmission modes of the data;

thirdly, on the basis of an Internet of things MQTT protocol, a middleware is used as a communication agent between an application end and nodes, an end-to-end management and control flow is designed when the nodes of the application end are found and deleted, the running state of the sensing nodes is monitored, and the working modes of the nodes are adjusted and controlled, data preprocessing and analysis allocation are added into the middleware, and transparent management and control of the nodes by an application layer are realized in cooperation with a communication agent mechanism;

fourthly, based on the data processing environment close computing of the Internet of things, data filtering computing is added into the middleware, and the network edge equipment can execute primary data computing and analyzing tasks.

2. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein a wireless sensor node resource virtualization management and control model comprises: the physical node resources are virtualized into virtual sensors convenient to process through software description in an abstract packaging mode, wherein an abstract model is divided into a fusion resource description model and an end-to-end management and control model, the fusion resource description model carries out centralized agreement on node attributes, data processing and storage, attributes with unique identification physical nodes and sensors carried by the nodes correspond to a data sensing process, and abstraction of hardware nodes is realized; the latter is the description of different levels of management and control modes of the application end, so that the application end integrates end-to-end management and control on different nodes, and in the two models, the middleware is used as a communication agent to be responsible for forwarding various data and management and control messages.

3. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein node digital identity fusion identification: the node digital identity + NetSort resource identification method is provided, wherein the node digital identity is the MAC address or IP address of a node when equipment leaves a factory, and NetSort is the communication network type of a certain node and marks the network communication type of the node;

4. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein node resource abstract description:

1. internet of things sensor node description model

The virtual sensor description model facing the multi-sensor improved XML, wherein the middleware encapsulates each attribute and data processing process of an entity sensor into a virtual sensor in an XML file form, and the virtual sensor comprises necessary sensor node deployment and use information, and comprises the following steps: 1) a node measurement type; 2) sensor metadata information; 3) converting and processing interface information by data flow; 4) a filtering interface of the data flow and an SQL query storage flow; 5) analyzing interface information of data edge; one or the same type of sensor resource is replaced by a virtual sensor, the middleware realizes sensor virtualization and dynamically processes the resource access process;

aiming at a middleware resource virtualization task and an end-to-end control task of an application end, inducing the content of metadata information of a sensor, wherein the metadata information comprises node resource digital identity, a unit of measurement data, a sensor observation type, a measurement range, a sampling period, a node running state and necessary time attributes;

2. virtual sensor abstraction pool

5. The multi-sensor internet of things resource adaptive deployment management middleware of claim 1, wherein based on end-to-end management of message brokers:

1. end-to-end resource management and control

Resource management and control are divided into four aspects: the method comprises the following steps that an application end manages and controls node resource deployment, node working mode management and control, node running state monitoring and edge data edge analysis and management and control based on environment proximity calculation, wherein the management and control in four aspects cover application layer management and control tasks, and the node resource deployment management and control refers to registration and deletion of node resources; the working mode management and control is that the application end manages and controls the node working mode according to the management and control instruction of the sensor; the monitoring of the node running state is to monitor whether the bidirectional communication between the node and the application layer in the sensing network fails; the data edge analysis means that the sensing data is directly subjected to basic analysis and calculation in a middleware, the whole management and control is carried out by an application end, the middleware is used as an auxiliary platform, the node working mode management and control, the node running state monitoring and the resource deployment management and control belong to the direct management and control of the application end, and the data edge analysis task is executed by the middleware;

2. dynamic publishing and subscribing model for message

3. resource management and control model based on gravity center

based on MQTT protocol, designing a message communication mechanism of an application end and a node based on gravity, calibrating node data sending and application end management and control information two types of Publish messages and Subcribe messages by using a set theme gravity, sending messages with specified gravity names for subscribing by an opposite side when the node needs to upload sensing data and the application end needs to issue a management and control command, and classifying message themes according to different transmission messages:

(1) node Publish message and application Subcribe message subject calibration with sensing data as content

(2) end-to-end management and control message subject calibration

6. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein an internet of things node discovery mechanism: after a set of nodes are designed in the form of theme messages and added into a perception network, an interactive process of authorized access is carried out when an application layer operates:

the second step: each node issues a registration message to the middleware after the connection is established, and the name of the center of gravity is Resg/Res digital identity;

the fourth step: after capturing that a node is added into a network, an application end verifies the Client digital identity, namely the Res digital identity in the message, if the digital identity is illegal or the application does not need to add the node, directly replying an unsubcribe message to a middleware, and directly ignoring the node by the middleware; if the digital identity is legal and the node to be accessed meets the requirement, the application end immediately issues an authorization message and receives the registration request of the node, the message subject is a ResgAck/Res digital identity, and the Res digital identity is the digital identity of a specific node allowed to be accessed;

and a sixth step: after the node is authorized to access, starting to push own sensing data;

7. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein an internet of things node deletion mechanism is: the application end actively initiates a process of deleting indication according to the change of the perception task or the abnormal perception of the node, and the process comprises three steps:

step 1: the application end firstly determines the node digital identity of the node to be deleted and sends a Publish message with the center of gravity of Delete/Res digital identity to the middleware;

and 3, step 3: after the middleware issues the subscription theme and is successfully matched, a certain node can successfully subscribe a deletion instruction, initiatively initiate disconnection of the Disconnect to the middleware to Disconnect message communication with the middleware;

8. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein a node running state monitoring process comprises: the application layer judges the specific node running state according to the data receiving condition and records the fault state in the equipment state table, the application end firstly detects whether the sensing data is received, when no data is received, the application end sends a Connect message in an MQTT protocol to determine whether the middleware can be connected, if the connection response CONNACK packet is not received, the middleware connection is interrupted, the middleware can be recovered after reconnection, and the connection response packet is not received, namely the middleware is failed;

by adopting the advice message, when the node and the middleware are disconnected suddenly, the application end finds the phenomenon that the node is disconnected suddenly in time;

after a node in a sensing network initiates connection to a middleware for the first time, a testament message is immediately sent to the middleware, the center of gravity of the message is named as Disconnected/Res digital identity, the application end subscribes the message with the center of gravity of Disconnected/#tothe middleware to receive the testament message of all nodes, the testament message is sent to the middleware only once, the message is stored by the middleware until the middleware finds that the node is Disconnected, the testament message is immediately sent to the application end which subscribes the message, and the application end determines that the sensing node with the node digital identity Res digital identity exits the connection due to abnormality.

9. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein a node working mode is adaptively adjusted:

1. sensor operating mode classification

the class 2 is related to task requirements, including data acquisition frequency, sensor attitude, restart or sleep time, and the management and control is related to decision of an application end and is suitable for all types of sensor nodes, and the management and control is convenient for the application end to reasonably plan the whole sensing network acquisition task according to specific requirements, energy consumption of a sensing layer and matching factors among nodes;

2. node working mode adjusting process

The working mode adjustment is a process that an application layer actively and perceptively initiates end-to-end communication with a network node, the transparent control is realized in that the application layer only needs to obtain Res digital identity of the node to be controlled, the control message is forwarded to a specific node under a specific network and is replaced by a middleware, and the working mode adjustment of the node needs to ensure that the operated node is in a communicable state:

10. The multi-sensor internet of things resource adaptive deployment management and control middleware of claim 1, wherein edge management and control analysis of data: based on the middleware, environment proximity calculation is adopted, and basic data management and control tasks are finished by the middleware, so that the management and control pressure and the data processing tasks of an application end are reduced, and the decision delay is reduced;