Wireless spiking neural networks (WSNNs) enable energy-efficient communications, especially in contexts of edge intelligence and learning, for both terrestrial (beyond 5G/6G) and Earth-space networking systems. Recent...
详细信息
ISBN:
(纸本)9798350351118;9798350351125
Wireless spiking neural networks (WSNNs) enable energy-efficient communications, especially in contexts of edge intelligence and learning, for both terrestrial (beyond 5G/6G) and Earth-space networking systems. Recent research work has revealed that distributed wireless SNNs (DWSNNs) achieve good performance in terms of inference accuracy and savvy use of energy in edge devices, under the constraints of limited bandwidth and spike loss probability. This technology appears thus promising for wireless sensor networks (WSNs) in space applications, where energy constraints are a major limiting factor. In this work, we focus on neuromorphic impulse radio (IR) transmission techniques for DWSNNs, considering different coding algorithms that implement IR modulations. Our quantitative evaluation relies on information-theoretic measures that should relate with different aspects of performance. Such measures may help achieving the optimal balance in DWSNNs applications, and their definition and evaluation in a simulated environment constitutes the main contribution of this manuscript.
暂无评论