RIS-Assisted CSIT-Free Data Fusion with Timing Misalignment

Over-the-air computation (AirComp) shows great potential in ultra-fast data fusion in Internet-of-Things (IoT) networks via exploiting the superposition property of wireless multiple access channels. However, to harvest the appealing advantages promised by Air-Comp, channel state information (CSI) at the transmitter side (CSIT) and strict synchronization among edge devices are required, which leads to high feedback overhead and expensive hardware cost in practical deployment. Therefore, how to achieve a low-overhead AirComp with only coarse-grained synchronization remains a question. To address this issue, we proposed a CSIT-free data fusion and timing resynchronization scheme for asynchronous AirComp systems with the aid of reconfigurable intelligent surface (RIS). Specifically, to overcome the reliance on CSIT and facilitate the low-overhead communication, we leverage the RIS to align the equivalent channel coefficients to the aggregation weights without the priori knowledge of CSI. Based on that, we further solve the timing misalignment problem by invoking a scaled matched filter, where a nominal delay that shifts back the weighted mean delay of the devices and a scaling factor are exploited for asynchrony elimination and noise mitigation purposes, respectively. Simulation results show that the proposed scheme provides excellent performances that approach the perfectly synchronized case.