The integration of the IIoT (Internet of Industrial Things) aims to improve the performance of the plant through the exploitation of the data collected by the various sensors/actors. The IIoT will make it possible to better know the plant, detect and respond to problems (panks, stock breaks) and manage to reduce or even eliminate downtime with machine learning on the basis of the data collected by the sensors. Among the applications of IIoT in the factory of the future are predictive maintenance, stock management and energy consumption management. The objective of this thesis is to propose algorithms derived from artificial intelligence as a means of decision-making regarding the processing of data generated by the objects of an IIoT network, depending on the level of criticality of the application, the nature of the object that generated the data and the state of the network. The actions taken by the various entities aim to make optimal use of the resources available (storage, calculation, bandwidth, radio resource..) to meet the needs of applications/objects. Game theory, multi-agent learning and negotiation are the tools we will use to make the IIoT network autonomous. Our proposals will be validated analytically, by simulation but also on the CPER PFEEXECL FFCA platform “Factories of Future Champagne-Ardenne” in which we are involved.