Semantic Communication Based Complexity Scalable Image Transmission System for Resource Constrained Devices

Multimedia traffic is expanding at an astonishing rate and is finding its way into many new applications, including wireless sensor networks and devices from the Internet of Things. However, the bandwidth and energy requirements associated with them are becoming increasingly prohibitive, challenging the sustainability of many communication networks. Semantic communications offer a novel approach to overcome bandwidth limitations, but the complexity of encoders and decoders limits their application in resources-constrained devices. We propose a semantic communication-based complexity-scalable image transmission system that uses asymmetric autoencoders to shift the complexity of the system to the encoder or decoder without compromising on human and machine perceived quality. Using a test data set, we demonstrate that the complexity can be shifted between the two without affecting the overall performance of the semantic communication system. This concept will be significant in the implementation of wireless sensor networks based on semantic communication and Internet of Things applications, while also providing a novel tool to further improve conventional encoder simplification approaches, such as distributed coding. Furthermore, the realization of energy efficient sensors by utilizing semantic communications and complexity scalability will have a direct impact on achieving sustainable development goals related to energy usage and management in media communications.