Multi-Slot Over-the-Air Computation in Fading Channels

IoT systems typically involve separate data collection and processing, face the scalability issue when number of nodes increases. For some tasks, only result fusion is needed. Then, whole process can be realized in an efficient way, integrating one step by over-the-air computation (AirComp). Its shortcoming, however, signal distortion channel gains are different, which cannot well solved transmission power control alone times deep fading. To address this issue, paper, we propose a multi-slot (MS-AirComp) framework for sum estimation fading channels. Compared with conventional (one slot each node) AirComp all nodes), MS-AirComp alternative policy that lies between them, exploiting multiple slots to improve so as facilitate control. Avoiding obtain instantaneous at sink key point. Specifically, transmissions distributed over threshold gain set scheduling. Each node transmits its once, first gets above threshold, or last remains below threshold. Theoretical analysis gives closed-form error channels, based on optimal parameters found. Noticing tends reduced cost more power, method suggested increase power. Simulations confirm proposed effectively reduce error, compared state-of-the-art methods.