EP3423906A1

EP3423906A1 - Smart node for a distributed mesh network

Info

Publication number: EP3423906A1
Application number: EP17712711.5A
Authority: EP
Inventors: Lionel BOUZON; Lionel DEBROUX; Charles EYNARD; Vincent DIMITRIOU; Christophe Voisin
Original assignee: Atos Worldgrid
Current assignee: Atos Worldgrid
Priority date: 2016-03-01
Filing date: 2017-03-01
Publication date: 2019-01-09
Also published as: CN109478056A; US20190155261A1; FR3048535A1; WO2017149050A1

Abstract

The invention relates to a smart node for a distributed mesh network, each node allowing two-way communication with other nodes or a central platform, and each node comprising a Linux or Linux-compatible computer hardware architecture and a software stack having a small footprint and low consumption of resources, using an object-oriented programming language to implement, by execution on the hardware architecture, the deployment of multiple nodes by neighbourhood and critical mass effect, each node being iso-functional and capable of communicating with its neighbour, storing and managing an object, and receiving an execute statement from a program of another node in the mesh, said program associating a node-specific identifier and a neighbourhood identifier with the communication module of each node.

Description

Smart node for distributed network according to a mesh

TECHNICAL FIELD OF THE INVENTION

The present invention relates to an intelligent node for distributed network according to a mesh and, more specifically to its use in the field of storage, distributed processing and the valuation of massive data.

BACKGROUND OF THE INVENTION

An industrial production chain usually involves several stages involving different actors or different entities. Each actor produces data and often these are managed locally. For example, in the case of the water treatment process and the valorization of the resources resulting from the treatment, a wastewater treatment plant locally pilot the industrial processes with a practice and a local driving experience at the considered site (monitoring, control). In terms of data, the site (factory) evolves in isolation with in particular no interface on the regional (territory) or global (networks and national resources).

The pooling of these different sources of information (local, regional and global sources) is, however, necessary because they would ensure the possibility of comparing data, their exchange and their valuation in order to reduce the footprint ecological and operating costs. The benefit of such an approach lies in the overall optimization of the industrial process and can be both environmental and economic. A solution to establish an interface between the different entities of the industrial process and a much more communication ecosystem It would be vast to use standard client / server architectures that use virtual private networks (VPNs).

However, setting up a VPN would lead to latencies and costs that are far too important as deployments could involve a multitude of different sites and entities.

The application TW201445929 (A) corresponding to the application EP2809034 teaches the use of nodes in an intelligent mesh network. The nodes then play the role of interface between the entities of the network. This paper teaches a method for establishing a cellular network architecture (SACM) mesh network by a method of: performing the deployment of a plurality of mesh nodes, wherein each of the mesh vertices has an ability to communicate with the other nodes of the mesh and a gateway capability for providing access to a service network, wherein the service network provides a wireless communication service; and consisting in establishing a plurality of links between the mesh nodes, each of the links connecting two of the mesh nodes; searching and selecting a plurality of dynamic gateway nodes from the plurality of mesh nodes and establishing a plurality of connections between the dynamic gateway nodes and the service network, and performing the routing and the path optimization to find the optimal route paths for all mesh nodes accessing the service network.

Only the vertices of the mesh have an ability to communicate with the other nodes of the mesh and the method requires a search and a selection of the nodes to perform the optimization of the mesh.

This requires a complex implementation and does not allow easy network scalability.

GENERAL DESCRIPTION OF THE INVENTION The purpose of the present invention is to overcome certain disadvantages of the prior art by proposing the capacity to interconnect a much larger data ecosystem allowing the consideration of an evolutionary dimension depending on the number of sites and also the type of data. applications.

This goal is achieved by an intelligent node for a network distributed according to a mesh, each node allowing bidirectional communication with other nodes or a central platform and each node comprising a hardware hardware architecture and a software stack, the node being characterized in that the execution of said software stack on the computer hardware architecture implements a set of functionalities comprising at least the following functionalities:

Creation and management, for itself or the other nodes of the mesh, of objects adapted to industrial processes so as to control any type of process, the set of objects defined for the whole mesh and known to each of the nodes being called "object dictionary";

Storing and managing at least one object by each node for maintaining the current state of the object and using a memorized list of neighborhoods of the nodes to which it is itself connected, to inform each neighboring node of the eventual change of state of the object;

Associating, with the communication module of each node, an identifier specific to the node and a neighborhood identifier, to make each node iso-functional and able to receive an execution command from a program of another node of the mesh; According to another particularity, the set of functionalities comprises the startup and communication functionalities of each node via a communication module, in its immediate vicinity thanks to a configuration with a minimum range of functionalities, the deployment of several nodes in distributed mesh. by critical mass and neighborhood effect and by optimizing, by means of an algorithm executed on the hardware architecture, the number of smart nodes to deploy and the number of their interconnections through neighborhoods to achieve availability, Robustness of deployment and continuity of service required by the quality of service of a specific service.

According to another particularity, the systematic distribution of data or commands is done by configuration of the broadcast module of the node (1) [Smart Node], this broadcasting module being initially configured to implement, by execution, on the hardware architecture computing, a function of diffusion, within the network (8), of a variable or a given group of variables with a given resolution and up to a given depth in the mesh, for example 3 or 4 levels of neighborhoods.

According to another particularity, the intelligent node comprises a node manager managing the dynamic deployment of new functionalities and functionalities implemented by the software modules, executed on the hardware hardware architecture, by monitoring and controlling the restart and updates. security day of a software module that would come to an end or die.

According to another particularity, each object belongs to at least one class which is a description of the characteristics of one or more objects representative of an industrial process or of a characteristic each object is created from this class and constitutes an instance of the class in question, the characteristics and state of an object are manipulated by methods incorporated into the intelligent node (1), the state of an object corresponds to the information stored at a given instant, as described by the values of all its properties, also called fields or attributes.

According to another particularity, each node comprises a device comprising at least one software layer, said software layer implementing, by execution on the computer hardware architecture, a storage function, in addition to the information coming from the process sensors, an attribute indicating that the concerned node is a parent of the object called "parent node". According to another particularity, each node comprises a device comprising at least one software layer, said software layer implementing, by execution on the computer hardware architecture, a function for informing each node (1) of its neighborhood so that the nodes (1) neighbors inform the other nodes (1) in a direction oriented in a direction that depends on the topology or architecture of the mesh, defining the links between the nodes of the network, and if necessary following a path oriented towards a platform (10). ) or to the processes (7a, 7b), each node (1) thus informing the rest of the mesh and each node (1) thus memorizing the object, its current state and the node (1) parent to which the object is assigned.

According to another particularity, the functionalities implemented by the node also include the broadcasting, in the form of time series, of the data collected or calculated by each node, said broadcasting being carried out by the association of two data dissemination modes: a so-called systematic diffusion mode in which the data is broadcast with a given resolution and up to a given depth in the mesh and, an opportunistic diffusion mode in which at least one neighboring node of another node concerned by a given initial request autonomously records the information or the data that passes through it on its memory in order to rebroadcast said data or information when a similar request to the original request is again raised.

According to another particularity, each node is configured to implement a planned functionality of systematic logging of the data, but also the storage of actions that take place periodically or actions relating to the so-called opportunistic mode of dissemination in its memory, each node thus having the capacity to behave autonomously for the historization of the data collected during the actions taking place periodically or actions relating to the so-called opportunistic mode of diffusion.

According to another particularity, each node has at least one access interface to its image of the "object dictionary", this interface being configured to define a new node or a new object for a node, the modification request being broadcast in the mesh and transmitted from one node to another until the parent node concerned if the change to the dictionary does not relate to the node from which the handler is accessed, the parent node of the object then proceeds to execution of the query, the result of the execution then being broadcast in turn in the rest of the mesh, each node receiving this result then updating its own image of the "object dictionary".

According to another particularity, the functionalities implemented by each intelligent node include the rebroadcast, via its broadcast module, to the rest of the mesh and at configurable time intervals the state of its own objects, in order to overcome any temporary break or persistent communication in the mesh, this capacity allowing the mesh of restore, if necessary, the integrity of the different images, associated with the different nodes of the mesh, of the "object dictionary"

According to another particularity, the objects are manipulated without the modifications made to the state of an object using the state of another object or influencing it, each object having an access authorization. for any use or any entity of the industrial process, the attributes or fields of definition of the objects being dynamically changed by the node manager.

According to another particularity, each object uses a method which defines a quality parameter associated with it, said quality parameter representing the difference between a desired target value of the state of an object and the actual state of the value, desired state of an object being formalized by the request to modify the state of said object, said request being formulated from any remote node even if it is not the parent node of the object, then transmitted to the mesh and from one node to another to the parent node concerned, the execution of said request by the concerned node thus allowing each node of the mesh to retrieve the value of the actual state of a object and thus to calculate its quality.

According to another particularity, each node of the mesh or the platform comprises a device comprising at least one software layer, said software layer implementing, by execution on a computer hardware architecture, a connection functionality to any node of the mesh, by sending the identifier of the node to be modified, so as to remotely and dynamically modify the node concerned even if the user is connected to a node that is not the parent node of the object he wants to modify.

According to another particularity, the nodes are used in an intelligent and universal system of industrial process supervision comprising a central mass data management platform for acquiring, managing and storing a data lake and communication means with a mesh distributed network consisting of intelligent nodes.

According to another particularity, each node performs the following functions:

• control of process monitoring sensors or programmable controllers for monitoring a process or actuators and acquiring data from them;

• data historization, formatting and decentralized calculations;

DESCRIPTION OF ILLUSTRATIVE FIGURES

Other features and advantages of the present invention will appear more clearly on reading the description below, made with reference to the accompanying drawings, in which:

FIG. 1 represents a diagram of the architecture of the smart node according to a first embodiment,

FIG. 2 represents an operating diagram of the smart node according to a second embodiment,

DESCRIPTION OF THE PREFERRED EMBODIMENTS

THE INVENTION The present invention relates to an intelligent node (1, FIG. 1) [Smart Node] for constituting with other nodes a network (8) distributed according to a mesh as represented in FIG.

In some embodiments, the intelligent node (1) has a hardware architecture and a low-resource (10-300 Mbyte), low resource-consuming software stack (2, 3, 4, 5, 6) running. on the hardware architecture. Said architecture being hardened (not shown), type X86 or ARM type Raspberry or MIPS, energy-saving and resistant to severe environmental conditions (shock and vibration, temperature -40 to + 80 ° C) operating under a system of LINUX operation or similar.

An intelligent node (1) [Smart Node] will conform to a micro services architecture. It has a backbone around which it is built: the Node Manager. Said node manager is programmed to activate dynamically and on demand a panel or set of features from the following categories:

Capacities of acquisition and instrumentation (example: in Modbus, or of Physical Signals)

Historization of data and messages, formatting and decentralized calculations on the data

Web dashboards for mobile operators

Gateway and Machine To Machine (M2M) Interoperability

In some embodiments, the systematic dissemination of data or commands is done by configuring the broadcast module of the node (1), this broadcast module being initially configured to implement, by execution on the hardware hardware architecture, a diffusion function, within the network (8), of a variable or a given group of variables with a given resolution and up to a given depth in the mesh, for example 3 or 4 neighborhood levels.

In some embodiments, additional functionality is first implemented as a module recognized by the node manager (12). Said node manager (12) thus manages the dynamic deployment of new functionalities and functionalities implemented by the software modules, executed on the hardware hardware architecture, by monitoring and controlling the restart and the security updates. a software module that would come to an end or die.

In some embodiments, each object created and managed belongs to at least one class that is a description of the characteristics of one or more objects representative of an industrial process, each object is created from this class and constitutes an instance of the class in question, the characteristics and the state of an object are manipulated by methods incorporated in the intelligent node (1) connected directly to the process whose objects are to be monitored and controlled The state of an object corresponds to the information stored at a given moment, as described by the values of all these properties, also called fields or attributes.

In some embodiments, each distributed network intelligent node (e.g., FIG. 2) includes middleware (2) incorporating a communications agent for communications bidirectional (1 1) with other nodes (1 b, 1 c, 1 d) or, preferably, to a central platform (10) and the neighboring node (1 d), to each neighboring node (1 e ) or (1f) as needed. This middleware (2) thus makes it possible to deploy a plurality of distributed network mesh nodes (8) by the critical mass and neighborhood effect of optimizing, by means of an algorithm executed on the hardware architecture, the number of intelligent nodes to deploy and the number of their interconnections through neighborhoods to achieve availability, robustness of deployment and continuity of service required by the quality of service of a specific service. Each node (1a, 1f) comprises a device comprising a device comprising at least one software layer, the execution of said software layer on the hardware hardware architecture implementing the storage (3) and management (4) functions. at least one object, ensuring the maintenance of the state of the object at each instant called "actual status" in English as opposed to "targeted status", and implementing at least one method of monitoring the change of the state of the object. This method uses a memorized list of the neighborhoods of the nodes (for example 1c, 1e, 1f) to which the node (for example 1d) is itself connected, to inform by the use of this list, each neighboring node of the eventual change of state of the object.

In some embodiments, said intelligent node for example (1a or 1f) comprises a device comprising at least one software layer (5, 4, 3), the execution of said software layer on the hardware hardware architecture enabling to memorize, thanks to a software layer (3) of logging, in addition to the information coming from the sensors of the process (7a or 7b respectively) and filling the fields of an object assigned to this node, an attribute representative of the identifier said node (1a or 1f respectively) and indicating that the node concerned is a parent of the object, said node being called "parent node". In some embodiments, the software stack (5, 6) of said node, as a result of processing by a processing engine (4), can send control signals to the sensors or the software layer (6). of programmable controllers or actuators for monitoring a process (7a or 7b) connected to the node (1a or 1f respectively) and by another software layer (5), the acquisition of the data coming from the sensors or actuators or programmable logic controllers processes (7a, 7b). Conversely, the processes (7a, 7b) will be able to trace the information to a platform (10) and then the data lake (9) through their respective neighboring nodes (1a, 1f).

In some embodiments, each node (for example 1f) informs each node of its neighborhood (for example 1 d) comprises a device comprising at least one software layer, said software layer implementing, by execution on the computer hardware architecture , a feature for informing each node (1) of its neighborhood so that neighboring nodes (1) inform the other nodes (1) in a direction oriented in a direction that depends on the topology or architecture of the mesh, defining the links between the nodes of the network, and if necessary following a path oriented towards a central platform (10) or to the processes (7a, 7b), each node (1) thus informing the rest of the mesh and each node (1) thus memorizing the object, its current state, and the parent node (1) to which the object is assigned.

Deployment of the nodes in the mesh (8) typically takes place as follows: the software of each node (1) is configured at the factory with a minimum range of functionality to start and communicate in its immediate vicinity. These basic functions enable a first diffusion of the characteristics of the node, namely its identifier and the configured objects. By object is meant any representation of business or technical data defining a variable and / or a service. In particular, we distinguish: - technical services: these are protocol service layers for communicating with industrial equipment (example: Modbus application layer, OPCUA, CanOPen, CAN);

- so-called primary variables: they are variables storing data collected by the technical services; - computing services: which allow the calculation of indicators and the implementation of logic based on the values of the primary variables;

- the secondary variables: which make it possible to store the results of the values of the computing services. It is therefore derived variables (by calculation) primary variables; - general services: these are various services (eg archiving, printing, taking photo or video, specific algorithm) to complete the traditional SCADA logic (Supervisory Control And Data Acquisition) by business or media services. For example, if a vibration indicator (secondary variable) exceeds a certain threshold then 10 photos are taken continuously by a general service and these photos are transmitted to the rest of the mesh.

The set of objects is gathered in the "dictionary of objects" The implementation of this dictionary, which is based on the basic use of NoSQL in memory, is innovative because very compact and independent of a pre-established object model . The compactness of the dictionary also allows diffusion throughout the mesh (8). Once the first commissioning of a node (1) established, it becomes visible and / or accessible from any other node (1) of the mesh (8). A local graphical interface incorporated in each node allows, then, to work on the easiest access node while allowing the configuration and modification of the services and functionalities of a remote intelligent node. The most common changes are:

- the addition of additional objects

- the modification of the data dissemination policy.

- the activation and / or implementation of an additional software module (example: the ability for the smart node to become a web server, the ability to inter-operate with a new system, etc.). An additional functionality is first implemented as a module recognized by the node manager [Node Manager] who will therefore be able to manage its life cycle. A node (1) [Smart Node] given is therefore configured to work with a certain panel of modules under the supervision and control of the node manager [Node Manager]. If a module breaks down or dies, the Node Manager is responsible for its restart and security updates.

One of the very innovative factors of the invention is that it is absolutely not necessary that the smart node [Node] on which a user works is the one he wishes to modify or be on the same network or in "IP visibility". direct "as would be a client with a classic server. Indeed the communication in the mesh rests on a communication of close by. For example and without limitation, a modification is made locally on a node A to a node E. The node A shares the same neighborhood (V _A , B) as the node B which has for neighborhood (V _B , A; VB, C; VB, D) - Node B shares the same neighborhood as Nodes C and D. Node D shares the same neighborhood as Node E. Node B is aware of the change made on Node A for the attention of E. As this one does not it does not concern him, he just informs his neighborhood, ie B, then B informs C and D, etc. Thus, step by step, the change request arrives in the neighborhood concerned and at the node [Smart Node] concerned (E). Node E proceeds with the change and if successful, it informs its neighborhood that its configuration has changed. The success notification of the modification is thus propagated back to the rest of the mesh.

In some embodiments, the dissemination of the data collected or calculated by the intelligent nodes (broadcast in the form of time series) is achieved by the association of two data dissemination modes: a so-called systematic broadcasting mode in which the data are broadcast with a given resolution and up to a given depth in the mesh and, a so-called opportunistic mode of diffusion in which at least one node (1) neighboring another node (1) concerned by a given initial request, saves autonomously the information or the data that passes through it on its memory in order to rebroadcast said data or information when a request similar to the initial request is again raised, the pattern or diagram of data broadcast broadcast by the nodes (1 ) being different from a systematic replication scheme in which the data distribution scheme is duplicated identically for all the nodes.

The systematic distribution is done by configuring the node's (1) [Smart Node] broadcast module. This broadcasting module is initially configured to broadcast a variable or a given group of variables with a certain resolution and up to a certain "depth". in the mesh "for example 3 or 4 levels of neighborhoods. This systematic policy allows, thus, at any point of the mesh to have a certain level of hypervision (not optimal with the best granularity and resolution, but a global vision all the same) The opportunistic diffusion is the capacity of the mesh to answer dynamically to a question asked. By For example, it may be necessary to draw finely, at the second between 14:03 and 14:08 from node A, the value of a pressure P1 acquired by the node E at intervals of half a second, but systematically diffused 20 minutes to the third and fourth level neighborhoods to which A. belongs. In this case, locally at node A, the data is not available so this node will form a specific request to try to recover the values. requested. The request will go step by step, from neighborhood to neighborhood until finding a node able to return a time series meeting the criterion. If the question is asked for the very first time, the answer will certainly be given by the node E. By broadcasting the answer, the so-called opportunistic policy consists for the other nodes of the mesh (8) to record on their cache memory (in the proxy sense -cache) this fraction of time series with a very precise resolution between 14:03 and 14:08. The next time the same question is asked again in the mesh (8) (this being very likely because it is certainly an epiphenomenon that may interest other users from other nodes) it will get a faster response because the answer will already be pre-stored by a neighbor node.

In some embodiments, the nodes are iso-functional, each node receiving an execution command from a program of another node of the mesh, the execution orders being either identical or different from one node. to another. Said program also associates with the communication module of each node an identifier specific to the node and a neighborhood identifier. The communication module having its node identifier and the neighborhood identifier, sends messages or requests to all the connections that it has had by wired or wireless means, for example and without limitation, if the request concerns the results of a measurement, it is sent to all nodes of the neighborhood. If among the nodes of the neighborhood there is at least one in the immediate vicinity of the sending node which has in memory the results, these the last are transferred to the node concerned. If the neighborhood nodes are unable to respond to the request, the request is forwarded to the nodes in their respective neighborhoods until the node that performed the measurements responds to them. If no node has made any measurements, the request is forwarded to the node in the vicinity of the sensor responsible for the measurements. Once these are done, the results are transmitted from neighborhood to neighbor to the node issuing the request.

In some embodiments, each node (1) is configured to implement a functionality, provided, systematic logging of data, but also the storage of actions that take place periodically or actions related to so-called opportunistic delivery mode in its memory, each node (1) thus having the ability to behave autonomously for the historization of the data collected during the actions taking place periodically or actions relating to the so-called opportunistic mode of diffusion.

The Middleware (2) (middleware) will have to be able to mesh mesh the nodes on separate and heterogeneous networks without having to resort to the construction of a Virtual Private Network (VPN). Deployments may involve a multitude of sites and different entities. The establishment of a VPN would lead to latency and costs far too important.

The middleware (2) will also need to implement security functions to ensure the confidentiality and integrity of the data exchanged between the nodes on the one hand and also with the brick of Big Data Management.

The middleware (2) will have to provide efficient two-way exchanges with the Big Data platform. It will initially have to support the data exchange then the distribution of pre-calculation services to be carried out at the nearest field or else the instances of implementation of prediction services (Machine Learning) previously trained on a central platform (10) of Big Data Management.

The dashboard restitution by the smart nodes of the mesh network is envisaged also as a thin client. At first they will also be generic and offer the possibility for the local user (factory or mobile) to plot the variables of his choice.

In some embodiments, the platform

central station (10) comprises a device comprising at least one software layer, said software layer implementing, by execution on the computer hardware architecture, the creation and management functionalities, for itself or itself or the other nodes (1) mesh, objects adapted to industrial processes to control any type of process. The set of objects defined for the entire mesh and known to each of the nodes is called "object dictionary".

In some embodiments, each node (1) has at least one access interface to an image of the "object dictionary", which interface is configured to define a new node or object for a node, the request for modification being broadcast in the mesh and transmitted from one node (1) to another until the parent node concerned if the modification made to the dictionary does not relate to the node from which the manager (12) is accessed, the node (1 ) parent of the object then proceeding, the execution of the query, the result of the execution then being broadcast in turn in the rest of the mesh, each node (1) receiving this result then updating its own image of the "object dictionary". This new architectural paradigm allows a real distribution of the supervision logic ensured collectively by the mesh by eliminating the use of a central single node. In some embodiments, the functionalities implemented by each intelligent node (1) comprise the rebroadcast, via its broadcast module, to the rest of the mesh and at configurable time intervals the state of its own objects, in order to mitigate to any temporary or persistent rupture of communication in the mesh, this capacity allowing the mesh to restore, if necessary, the integrity of the different images, associated with the different nodes of the mesh, of the "object dictionary".

In some embodiments, the objects are manipulated without the changes made to the state of an object using the state of another object or influencing it, each object having a permission of access for any use or entity of the industrial process, the attributes or definition fields of the objects being dynamically changed by the node manager (12).

In some embodiments, each object uses a method that defines a quality parameter associated therewith, said quality parameter representing the difference between a desired target value of the state of an object and the actual state of the value, the desired state of an object being formalized by the request to modify the state of said object, said request being formulated from any remote node (1) even if it is not the parent node of the object 'object, then transmitted to the mesh and from one node (1) to another to the node (1) concerned parent, the execution of said request by the node (1) concerned thus allowing each of the nodes ( 1) of the mesh to recover the value of the real state of an object and thus to calculate its quality.

In some embodiments, the nodes (1) may be used in an intelligent and universal industrial process supervision system preferably comprising a central mass data management platform (10) (for example and without limitation , Big Data Management) for acquiring, managing and memorizing a lake (9, FIG. 2) and communication means (1 1) with a network (8) distributed in mesh consisting of intelligent nodes (1 a to 1 f). The structuring of the supervisory logic in what is called the object dictionary allows us to define, with the help of "atomic parts" (mainly services and object variables), what is generally defined as a monolithic application in the field. state of the art industrial supervision systems. All services form what is called "supervisory intelligence". The latter is made "transportable" thanks to the middleware (2), able to distribute both data and "intelligence" on different nodes (1) of the network (8) thus forming the supervision network. Embedded, local service engines are able to handle services of different types (not just computations) in a homogeneous way and regardless of the hardware platform. The main advantage of the engines in charge of the services is that they can operate on small hardware units (low CPU resources [Central processing unit or CPU] and memory).

The mesh network of intelligent nodes makes it possible to deploy an infrastructure of industrial supervision, simulation or collection and data analysis particularly adapted to physically very distributed processes (micro power plants, loT, distribution, wind or tidal fields, open-field sensors , process simulation).

The mesh network of intelligent nodes is an innovative software solution in the field of industrial supervision. Its aim is to complement the Supervisory Control And Data Acquisition (SCADA) of the market to enable an agile and rapid deployment of permanent or temporary supervision schemes.

Unlike traditional SCADAs based on central and very vertical server-type systems, the mesh network of intelligent nodes makes it possible to define a supervision strategy carried by a set of software nodes connected by intelligent middleware. It is It is therefore possible to define a supervision logic that is as close as possible to the process and with very fine granularity. This granularity makes it possible, among other things, to be able to define individual permissions on each of the elements of the supervision (variables, algorithms) making the meshed network of intelligent nodes a multi-user system but especially multi-entity. Atomic manipulation of the supervisory elements also enables hot monitoring deployments and scaling while limiting the risk of regression on existing logic. These software nodes can operate on a wide range of hardware including mobile devices (smartphones and tablets) but also on-board industrial field equipment. The mesh network of intelligent nodes thus introduces the mobility in a native way while proposing to each user, whatever his point of connection, the same quality of global hypervision as for a central system.

Finally, the mesh network of intelligent nodes offers a wide range of remote deployment features on the most constrained network topologies (complex routing, reduced bandwidth) and without the need to use VPN. These innovations allow time and role decorrelation of maintenance and scaling actions by eliminating the need for computer skills, PLC (Programmable Logic Controller or SCADA) in the field. As such, the mesh network of smart nodes is a Plug and Play solution where the operational ones limit themselves to connect the devices (devices) of ground to the electrical network and the communication network. The rest of the deployment (software bricks and supervision logic) is then provided by remote users. The mesh network of intelligent nodes is therefore particularly suitable for physically distributed industrial processes and where the maintenance and scaling costs are an important factor. Nevertheless in a factory supervision context, the mesh network of intelligent nodes also brings many advantages.

In the context of the renovation or evolution of existing industrial supervision the addition of software nodes, the mesh network of smart nodes quickly allows to set up advanced mobility functions throughout the perimeter of the plant. Mesh construction of the solution also allows a progressive open deployment to changes in the strategy based on the first user feedback. Many features can also be used for Smart Maintenance or predictive maintenance. Complementary instrumentations (sensors, concentrator) and analysis (traces, history, alarms, dashboards) can be created on the fly and in active collaboration between several users for the purpose of process optimization or implementation of asset tracking (electronic documentation, QRCodes).

The mesh network of smart nodes seen as a dynamic tool also allows the deployment of temporary supervision and analysis strategies that are very adapted to the audit phases (energy efficiency, simulation, security and industrial safety). For this type of deployment in particular, the mesh network of intelligent nodes is compatible with a wide range of wireless communication in the ISM band (169Mhz, 868Mhz, 969Mhz) and in particular with the technology SIGFOX (Ultra Narrow Band) and LoRA. In certain embodiments, each node (1) of the mesh or the central platform (10) comprises a device comprising at least one software layer, said software layer implementing, by execution on a computer hardware architecture, a connection functionality to any node (1) of the mesh (8), by sending the identifier of the node to be modified, so as to remotely and dynamically modify the node (1) concerned even if the user is connected to a node ( 1) which is not the parent node of the object it wants to modify.

Thus the solution developed is able to acquire data from internal treatment processes at the plant or upstream and downstream thereof in the natural environment.

On the other hand it is also able to integrate other data related to the data ecosystem and relevant for the intelligent and end-to-end driving of it. These are referred to as external sources in the rest of this document (example: weather, trading, energy markets).

Data sources are therefore very different in nature. The need to know how to interfere with specialized sensors in open field (SigFox or LoRA wireless link for example), compatibility with a panel of instrumentation protocols or industrial interoperability (example: Modbus, OPC, OPCUA), compatibility with interoperability standards for moving objects or people (M2M ETSI) and finally the possibility of querying third-party systems on the cloud (example: Web content analysis and exploitation, REST API, databases). hydrological data, flood management for example in the case of a water treatment plant).

Thus, the software layer (6) for sending the control signals and the software layer (5) for acquiring the data, to or from the sensors or actuators or programmable logic controllers of the processes, must cooperate with a hardware allowing connections without wire in addition to links wired and compatible with a panel of instrumentation or interoperability protocols for objects or people in motion.

The mesh deployment of the network (8) makes it possible to interface with the data ecosystem at various aggregation levels (terrain, vehicles, factories, regional area, global level or cloud). An architecture of this type thus makes it possible to collect the data at the most adapted levels. The transmission of the data can therefore involve several nodes before provision of the central platform (10) of Big Data Management (BDM) for its valuation.

After valorization and creation, for example of relevant indicators

(fields of an object) for the conduct of the process, the inter-node communication can be used again but "in the direction of the descent" this time to transmit optimized instructions to the field. Here again the mesh architecture makes it possible to overcome direct communication links between the global level and the field.

The implementation of the functional layers of this global information system therefore operates at different levels of aggregation, so it is not only a classical central system but an infrastructure capable of addressing a problematic problem. highly distributed (sensors in the wild, vehicles or people on the move, factories on different sites, interoperable with third-party regional systems or central systems) and highly dynamic (deployments of new services, scalability, adaptability to availability or not of parameters necessary for the hyper-vision or optimization of driving processes.

The present application describes various technical features and advantages with reference to the figures and / or various embodiments. Those skilled in the art will appreciate that the technical features of a given embodiment may in fact be combined with features of another embodiment unless the reverse is explicitly mentioned or it is evident that these features are incompatible or that the combination does not provide a solution to at least one of the technical problems mentioned in this application. In addition, the technical features described in a given embodiment can be isolated from the other features of this mode unless the opposite is explicitly mentioned.

It should be obvious to those skilled in the art that the present invention allows embodiments in many other specific forms without departing from the scope of the invention as claimed. Therefore, the present embodiments should be considered by way of illustration, but may be modified within the scope defined by the scope of the appended claims, and the invention should not be limited to the details given above.

Claims

1. An intelligent node (1) for a network (8) distributed according to a mesh, each node (1) allowing bidirectional communication (1 1) with other nodes (1) or a central platform (10) and each node comprising a hardware architecture computer and a software stack (2, 3, 4, 5, 6) using an object-oriented language, the node being characterized in that the execution of said software stack on the hardware hardware architecture implements a set of functionalities comprising at least the following features:

• creation and management, for itself or the other nodes (1) of the mesh, objects adapted to industrial processes so as to control any type of process, the set of objects defined for the whole mesh and known each of the nodes being called "object dictionary";

Storing and managing at least one object by each node (1) for maintaining the current state of the object and using a memorized list of neighborhoods of the nodes to which it is itself connected, to inform each node ( 1) next to the eventual change of state of the object;

• associating, with the communication module of each node (1), an identifier specific to the node and a neighborhood identifier, to make each node iso-functional and able to receive an execution command from a program of another node (1) of the mesh;

2. Smart node (1) according to claim 1, characterized in that the set of functionalities comprises the startup and communication functions of each node (1) via a module of communication, in its immediate vicinity through a configuration with a minimum range of features, deployment of multiple nodes (1) in mesh (8) distributed by critical mass effect and neighborhood and consisting of optimizing, by means of an algorithm executed on the hardware architecture, the number of intelligent nodes (1) to deploy and the number of their interconnections through neighborhoods to achieve availability, robustness of deployment and continuity of service required by the quality of service of a specific service.

3. intelligent node (1) according to claim 1, characterized in that the systematic dissemination of data or commands is done by configuration of the broadcast module node (1) [Smart Node], the broadcast module being initially configured for implement, by execution on the computer hardware architecture, a broadcasting function, within the network (8), of a variable or of a given group of variables with a given resolution and up to a given depth in the mesh, for example 3 or 4 levels of neighborhoods.

4. intelligent node (1) according to claim 1, characterized in that it comprises a node manager (12) managing the dynamic deployment of new functionalities and functionalities implemented by the software modules, executed on the hardware architecture computer, by monitoring and controlling the restart and security updates of a software module that would come to a halt or die.

5. intelligent node (1) according to claim 1, characterized in that each object belongs to at least one class which is a description of the characteristics of one or more objects representative of an industrial process or a business characteristic, each object being created from this class and constituting an instance of the class in question, the characteristics and the state of an object being manipulated by methods incorporated into the intelligent node (1) the state of a corresponding object stored information at a given moment, such as described by the values of all its properties, also called fields or attributes.

6. intelligent node (1) according to claim 1, characterized in that each node (1) comprises a device comprising at least one software layer, said software layer implementing, by execution on the computer hardware architecture, a functionality of storing, in addition to the information from the process sensors (7a, 7b), an attribute indicating that the concerned node (1) is a parent of the object called "parent node".

7. intelligent node (1) according to claim 1, characterized in that each node (1) comprises a device comprising at least one software layer, said software layer implementing, by execution on the computer hardware architecture, a functionality to inform each node (1) of its neighborhood so that neighboring nodes (1) inform the other nodes (1) following a path oriented in a direction that depends on the topology or architecture of the mesh, defining the links between the nodes of the network , and if necessary following a path oriented towards a central platform (10) or to the processes (7a, 7b), each node (1) thus informing the rest of the mesh and each node (1) thus memorizing the object, its state current node and the parent node (1) to which the object is assigned.

8. Smart nodes (1) according to one of the preceding claims, characterized in that said functionalities implemented by the node (1) also comprises the diffusion, in the form of time series, data collected or calculated by each node (1) said diffusion being carried out by the association of two data broadcasting modes: a so-called systematic broadcasting mode in which the data are broadcast with a given resolution and up to a given depth in the mesh and, a so-called diffusion mode in which at least one node (1) neighboring another node (1) concerned by a given initial request, records in a manner autonomous information or data that passes through it on its memory in order to rebroadcast said data or information when a request similar to the initial request is again raised, the pattern or diagram of data broadcast broadcast by the nodes (1) being different from a systematic replication scheme in which the data distribution scheme is duplicated identically for all nodes.

9. Smart nodes (1) according to one of the preceding claims, characterized in that each node (1) is configured to implement a functionality, provided systematic data historization, but also the storage of actions that take place periodically or actions relating to the so-called opportunistic mode of dissemination in its memory, each node (1) thus having the ability to behave autonomously for the historization of the data collected during the actions taking place periodically or actions relating to the mode of diffusion says opportunists.

10. Smart nodes (1) according to one of the preceding claims, characterized in that each node (1) has at least one access interface to an image of "object dictionary", this interface being configured to define a new node or a new object for a node, the modification request being broadcast in the mesh and transmitted from one node (1) to another to the parent node concerned if the modification made to the dictionary does not relate to the node from which the manager (12) is accessed, the node (1) parent of the object proceeding then, the execution of the query, the result of the execution then being broadcast in turn in the rest of the mesh, each node (1 ) receiving this result then updating its own image of the "object dictionary".

1 1. Smart nodes (1) according to one of the preceding claims, characterized in that said functionalities implemented by each intelligent node (1) comprises the rebroadcast, via its broadcast module, to the the remainder of the mesh and at configurable time intervals the state of its own objects, in order to overcome any temporary or persistent rupture of communication in the mesh, this capacity allowing the mesh to restore, if necessary, the integrity of the different images, associated with the different nodes of the mesh, of the "object dictionary".

12. Smart nodes (1) according to one of the preceding claims, characterized in that the objects are manipulated without the changes made to the state of an object do not call the state of another object nor influence the latter, each object having an access authorization for any use or any entity of the industrial process, the attributes or definition fields of the objects being dynamically changed by the node manager (12).

13. intelligent nodes (1) according to one of the preceding claims, characterized in that each object uses a method that defines a quality parameter associated with it, said quality parameter representing the difference between a desired target value of the state of an object and the real state of the value, the desired state of an object being formalized by the request to modify the state of said object, said request being formulated from any remote node (1) even it is not the parent node of the object, then transmitted to the mesh and from one node (1) to another to the parent node (1) concerned, the execution of said request by the node (1 ) concerned thus allowing each of the nodes (1) of the mesh to recover the value of the real state of an object and thus to calculate its quality.

14. Smart nodes (1) according to one of the preceding claims, characterized in that each node (1) of the mesh or the central platform

(10) comprises a device comprising at least one software layer, said software layer implementing, by execution on a computer hardware architecture, a connection functionality to any node (1) of the mesh (8), by sending the identifier of the node to be modified, so as to remotely and dynamically modify the node (1) concerned even if the user is connected to a node (1) that is not the parent node of the object he wants to modify.

15. Set of smart nodes (1) according to claim 1, characterized in that they are used in an intelligent and universal industrial process supervision system comprising a platform (10) central mass data management for acquisition management and storage of a lake (9) of data and means of communication with a network (8) distributed in mesh formed of intelligent nodes (1).

Smart node (1) according to claim 1, characterized in that each node performs the following functions:

Monitoring process monitoring sensors (7) or programmable controllers (7) for monitoring a process or actuators and acquiring data therefrom;

• data historization, formatting and decentralized calculations.