METHOD OF ASSOCIATION OF USER EQUIPMENT IN A CELLULAR NETWORK BY MEANS OF MULTI-AGENT REINFORCEMENT LEARNING

摘要

The present invention relates to a method of associating user equipment with base stations of a cellular network, in particular of a heterogeneous network such as a 5G network, said method using a learning algorithm by multi-reinforcement. -agent (MARL). An agent associated with user equipment receives an observation of its environment and derives an action from it, this action resulting in a request for association of the user with a neighboring base station. This action is chosen according to a strategy of the user seeking the maximum of an action value function for said observation and a plurality of possible actions, the action value function being defined as the expectation of the sum the agent's future rewards discounted by a discount factor. Once the actions have been performed by the different users, a common reward is provided by the network to the different users, this reward being zero in the event of a collision of association requests and the result of a utility function otherwise. The action value function for the different possible actions is predicted by a recurrent neural network (DRQN) trained on a set of experiences stored in local memory. A variant based on a strategy distillation is also proposed to allow a better adaptation of the association strategy to different possible tasks of the network. Figure for the abstract: Fig. 2