Mengyang Wang
Mengyang Wang has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Jiahui Li
School of Com- puter Science Wireless Technology Lab Harbin Institute of Technology 150001 Harbin China
2 shared publications - Mengyao Ma
Wireless Technology Lab Huawei 518129 Shenzhen China
2 shared publications - Xiaopeng Fan
Harbin Institute of Technology
2 shared publications - Yonghong Tian
Peking University
1 shared publication
Latest Publications
- Open Access
Spiking Semantic Communication for Feature Transmission with HARQ
In Collaborative Intelligence (CI), the Artificial Intelligence (AI) model is divided between the edge and the cloud, with intermediate features being sent from the edge to the cloud for inference. Several deep learning-based Semantic Communication (SC) models have been proposed to reduce feature transmission overhead and mitigate channel noise interference. Previous research has demonstrated that Spiking Neural Network (SNN)-based SC models exhibit greater robustness on digital channels compared to Deep Neural Network (DNN)-based SC models. However, the existing SNN-based SC models require fixed time steps, resulting in fixed transmission bandwidths that cannot be adaptively adjusted based on channel conditions. To address this issue, this paper introduces a novel SC model called SNN-SC-HARQ, which combines the SNN-based SC model with the Hybrid Automatic Repeat Request (HARQ) mechanism. SNN-SC-HARQ comprises an SNN-based SC model that supports the transmission of features at varying bandwidths, along with a policy model that determines the appropriate bandwidth. Experimental results show that SNN-SC-HARQ can dynamically adjust the bandwidth according to the channel conditions without performance loss.
Open AccessSpiking Semantic Communication for Feature Transmission with HARQ
In Collaborative Intelligence (CI), the Artificial Intelligence (AI) model is divided between the edge and the cloud, with intermediate features being sent from the edge to the cloud for inference. Several deep learning-based Semantic Communication (SC) models have been proposed to reduce feature transmission overhead and mitigate channel noise interference. Previous research has demonstrated that Spiking Neural Network (SNN)-based SC models exhibit greater robustness on digital channels compared to Deep Neural Network (DNN)-based SC models. However, the existing SNN-based SC models require fixed time steps, resulting in fixed transmission bandwidths that cannot be adaptively adjusted based on channel conditions. To address this issue, this paper introduces a novel SC model called SNN-SC-HARQ, which combines the SNN-based SC model with the Hybrid Automatic Repeat Request (HARQ) mechanism. SNN-SC-HARQ comprises an SNN-based SC model that supports the transmission of features at varying bandwidths, along with a policy model that determines the appropriate bandwidth. Experimental results show that SNN-SC-HARQ can dynamically adjust the bandwidth according to the channel conditions without performance loss.
- Open Access
SNN-SC: A Spiking Semantic Communication Framework for Feature Transmission
In Collaborative Intelligence (CI), Artificial Intelligence (AI) models are split between edge devices and cloud. Features extracted from input on edge devices are transmitted to the cloud for subsequent tasks. Extracting task-related and compact information is critical when transmission bandwidth is limited. In this paper, we propose a task-oriented Semantic Communication (SC) framework (SNN-SC) to address this problem. In SC, only important information for downstream tasks is extracted and transmitted. However, most of the existing SC works only transmit analog information on the AWGN channel and cannot be directly used for digital channels. SNN-SC fills this gap by using Spiking Neural Networks (SNNs) to extract and transmit semantic information. Since the outputs of SNNs are binary spikes, SNN-SC can be directly applied to digital channels. In SNN-SC, a new spiking neuron is proposed to help the cloud recover binary semantic information into informative floating-point features. Furthermore, we improve the performance of SNN-SC by maximizing the entropy of the semantic information. We evaluate the performance of SNN-SC on different collaborative classification models, digital channels, and bandwidths. The experimental results show that SNN-SC is more robust than the CNN-based SC framework and separate source and channel coding method on digital channels.
Open AccessSNN-SC: A Spiking Semantic Communication Framework for Feature Transmission
In Collaborative Intelligence (CI), Artificial Intelligence (AI) models are split between edge devices and cloud. Features extracted from input on edge devices are transmitted to the cloud for subsequent tasks. Extracting task-related and compact information is critical when transmission bandwidth is limited. In this paper, we propose a task-oriented Semantic Communication (SC) framework (SNN-SC) to address this problem. In SC, only important information for downstream tasks is extracted and transmitted. However, most of the existing SC works only transmit analog information on the AWGN channel and cannot be directly used for digital channels. SNN-SC fills this gap by using Spiking Neural Networks (SNNs) to extract and transmit semantic information. Since the outputs of SNNs are binary spikes, SNN-SC can be directly applied to digital channels. In SNN-SC, a new spiking neuron is proposed to help the cloud recover binary semantic information into informative floating-point features. Furthermore, we improve the performance of SNN-SC by maximizing the entropy of the semantic information. We evaluate the performance of SNN-SC on different collaborative classification models, digital channels, and bandwidths. The experimental results show that SNN-SC is more robust than the CNN-based SC framework and separate source and channel coding method on digital channels.