NEURONAS funded by HRFI

Reconfigurable intelligent surfaces (RISs) operate like electroma-gnetic (EM) mirrors that go beyond the Snell law to create a beneficial EM environment that provides flexible adaptation creating sustainability in terms of energy efficiency & economic de-ployment. However, they cannot operate in realistic conditions, i.e. with moving user equipment, due to the high latency of their control units. Specifically, they use low-complexity microcontrol-lers that can reconfigure in a frequency of some MHz to ensure low-computation power con-sumption. Of note the coherence time of millimeter wave channels is in the orders of μs.

The goal of NEURONAS is to design & develop a new type of RIS that support high mobility through real-time adaptation to the ever-changing wireless channel conditions. In this direction, the neuromorphic processing unit (NPU) coordinates all the RIS units, to create a beneficial EM environment. In this direction, a pro-active approach that predicts the user & possible blockers future positions need to be created. The approach is event-driven; thus, spike neural networks (SNNs) and/or spike reinforcement learning (SRL) are used. 

NEURONAS has the following strategical objectives: i) To create a new type of RIS by replacing the conventional microcontroller with a NPU; ii)To design and use meta-materials that reduce even more the unit cell response time (from 2-4 ns to 1-3 ns); iii) To design & assess SNNs & SRLs based methods for proactive beam tracking; iv) To design, develop & assess SRLs for real time adaptation; and v) To demonstrate the functionalities in realistic conditions.