Ph. D. Project
Dynamic reconfiguration of a scalable network of communicating objects, under service and energy
constraints, by simulation
Dates:
2023/09/15 - 2026/09/16
Student:
Supervisor(s):
Description:
dvances in the miniaturization of micro-electronic systems, coupled with those in the field of
embedded networks, make it possible to design «smart» objects, whose intrinsic matter itself forms a
network of communicating elements. An object or a set of this kind of objects make it possible to
exploit all the promises of the Internet of Things (IoT). Dynamic reconfiguration (by fusion or
disintegration) of the network formed by such objects requires the definition of an appropriate control
system with the aim of minimizing the energy consumption of objects while guaranteeing their
respective information collection services.
A network of communicating elements is modelled by a related graph G = (N, L) where N represents
the set of Nodes of the graph and L is the set of non-oriented Links between nodes. Recovery energy
techniques and development of low-energy or energy-efficient communication protocols help to limit
energy consumption and extend the life of a static network of communicating objects [Rault 2015; ...].
Similarly, various decentralized network organization strategies have been proposed to optimize the
data collection [Delgado 2014]. Generally, the questioning of the communication structure is locally
initiated by the network nodes. The ambition is to "manage the collection of information" and the
network structure by an external controller according to the principles of "edge-computing" and the
"Software Defined Networking" paradigm. Through this work, it is a matter of prolonging the open-
loop optimization of the physical part, by an online simulation in the digital space allowing to decide
whether to reorganize the physical part. The modeling of decision-making processes can be done by
applying work concerning the management of manufacturing systems [Miradamadi 2009], which very
often use the concept of Multi-Agent Systems [Hoang 2012].
The simulation allows to predict the future amount energy of objects according to the current
communication structure and to consider the questioning of this structure by continually testing other
candidate solutions. Because the context of object can change (object lifecycle and therefore of service
change, or encounter other objects, or because some nodes have little or no more energy), it will be
necessary to update the communication structure and therefore proceed to the organization phase
again. Thus, two distinct periods are to be considered for the energy prediction of the network of
objects:
- a collection period during which the network cyclically pulls the data up to a collector node for
external processing (i.e. E, Coll, the energy consumed during a collection phase). This phase will be
repeated until the controller decides to reorganize the WSN.
- An organisation period during which a new structure defined by the on-line simulation is established
(EOrg, the energy consumed during the implementation phase of the new organisation);
Designing the generic «middleware» allowing the monitoring of a scalable network of communicating
objects is the main challenge. The following sub-issues will need to be addressed (non-exhaustive list):
a. Represent the digital twin and simulate a network of communicating objects,
b. Define and model the control of a static network based on the on-line simulation,
c. Drive data collection from a scalable network in an optimal manner.
The main scope of this work will be production and logistics systems in the field of either
manufacturing or construction. This work can be applied to the concept of «communicating matter»
(MC) developed several years ago at CRAN [Kubler 2012, Mekki 2016, Wan&al. 2020]. Through the
ANR McBIM (Communicating Material for BIM) project, led by the CRAN, the optimization of physical
data collection was studied in the work of LAAS [Loubet&al. 2018] and CRAN [Wan&al. 2019; Wan&al.
2020], without taking into account the "services" aspect. Thus, this thesis would incorporate this
notion based on the work developed in the McBIM project.
embedded networks, make it possible to design «smart» objects, whose intrinsic matter itself forms a
network of communicating elements. An object or a set of this kind of objects make it possible to
exploit all the promises of the Internet of Things (IoT). Dynamic reconfiguration (by fusion or
disintegration) of the network formed by such objects requires the definition of an appropriate control
system with the aim of minimizing the energy consumption of objects while guaranteeing their
respective information collection services.
A network of communicating elements is modelled by a related graph G = (N, L) where N represents
the set of Nodes of the graph and L is the set of non-oriented Links between nodes. Recovery energy
techniques and development of low-energy or energy-efficient communication protocols help to limit
energy consumption and extend the life of a static network of communicating objects [Rault 2015; ...].
Similarly, various decentralized network organization strategies have been proposed to optimize the
data collection [Delgado 2014]. Generally, the questioning of the communication structure is locally
initiated by the network nodes. The ambition is to "manage the collection of information" and the
network structure by an external controller according to the principles of "edge-computing" and the
"Software Defined Networking" paradigm. Through this work, it is a matter of prolonging the open-
loop optimization of the physical part, by an online simulation in the digital space allowing to decide
whether to reorganize the physical part. The modeling of decision-making processes can be done by
applying work concerning the management of manufacturing systems [Miradamadi 2009], which very
often use the concept of Multi-Agent Systems [Hoang 2012].
The simulation allows to predict the future amount energy of objects according to the current
communication structure and to consider the questioning of this structure by continually testing other
candidate solutions. Because the context of object can change (object lifecycle and therefore of service
change, or encounter other objects, or because some nodes have little or no more energy), it will be
necessary to update the communication structure and therefore proceed to the organization phase
again. Thus, two distinct periods are to be considered for the energy prediction of the network of
objects:
- a collection period during which the network cyclically pulls the data up to a collector node for
external processing (i.e. E, Coll, the energy consumed during a collection phase). This phase will be
repeated until the controller decides to reorganize the WSN.
- An organisation period during which a new structure defined by the on-line simulation is established
(EOrg, the energy consumed during the implementation phase of the new organisation);
Designing the generic «middleware» allowing the monitoring of a scalable network of communicating
objects is the main challenge. The following sub-issues will need to be addressed (non-exhaustive list):
a. Represent the digital twin and simulate a network of communicating objects,
b. Define and model the control of a static network based on the on-line simulation,
c. Drive data collection from a scalable network in an optimal manner.
The main scope of this work will be production and logistics systems in the field of either
manufacturing or construction. This work can be applied to the concept of «communicating matter»
(MC) developed several years ago at CRAN [Kubler 2012, Mekki 2016, Wan&al. 2020]. Through the
ANR McBIM (Communicating Material for BIM) project, led by the CRAN, the optimization of physical
data collection was studied in the work of LAAS [Loubet&al. 2018] and CRAN [Wan&al. 2019; Wan&al.
2020], without taking into account the "services" aspect. Thus, this thesis would incorporate this
notion based on the work developed in the McBIM project.
Keywords:
Digital Twin, IoT, Network Control, Energy Efficiency, Online Simulation, Mulit-Agent System
Department(s):
Modeling and Control of Industrial Systems |