Apprentissage et adaptation au canal de propagation pour liens radiofréquence ultra-faible consommation dédiés aux applications à très grande autonomie

The widespread adoption of applications based on the Internet of Things (IoT) requires a considerable increase in the lifetime of communicating objects to be energy independent for very long periods (greater than 10 years). A possible track to reduce the average consumption of a node is to provide the transceiver radio frequency (RF), the node component that is often the most power hungry, the intelligence that it adapts its performance to the dynamic conditions RF propagation channel, thereby avoiding unnecessary dissipate energy when propagation conditions are good. Indeed, given the very large dynamic signal observed in all real RF system, typically of the order of 30 to 100 dB, the potential energy savings deposits are very considerable. The heart of adaptive RF system is the mechanism for estimating the quality of the link. Although historically the link quality indicators based on the level of the received signal or the frame error rate were unreliable and difficult to use, new link quality indicators are emerging in the literature, indicators able to reconstruct an image the signal to noise ratio (SNR) of the received symbol. We postulate that smart transceivers future will exploit to extract specific information on propagation conditions and, where applicable, the likely reasons for the deterioration of the quality of the link (masking, fading, interference due to simultaneous RF transmissions ( co-channel, adjacent, within and between networks), RF interference due to various electromagnetic noise in the environment (microwave ovens), etc ...) to adapt and minimize energy consumption.