Integrating of GAI-Based Consumer IoV and Multiple Access: A Novel Bundled Approach

Generative artificial intelligence (GAI) has attracted widespread attention since its creative artificial intelligence-generated content (AIGC) can be produced to meet the requirements of consumer electronics. In particular, bridging GAI and consumer Internet of Vehicles (IoV) is promising to serve mobile vehicles with complicated data processing, environmental perception, decision-making, and autonomous navigation. On the other hand, multiple access is one of the key approach to deal with ever-increasing data, benefited from its superior spectrum efficiency. However, the integrating of multiple access with GAI-based consumer IoV networks is still difficult to achieve, since the terrible interferences between vehicle devices. To solve this dilemma, taking uplink GAI-based consumer IoV as an example, we investigate a novel bundled scheme to improve received signal-to-noise ratio (SNR) and prevent outage events effectively, developing bundled non-orthogonal multiple access (B-NOMA), bundled rate-splitting multiple access (B-RSMA), and enhanced B-RSMA (EB-RSMA). Meanwhile, our proposed bundled schemes also solve the problem that the interference power exceeds the desired signal resulting in a significant performance loss, regarding both NOMA and RSMA decoding principles. In addition, since that our proposed schemes firmly depend on accurate channel state information (CSI), a GAI-based deep neural network (DNN) technique is further proposed for channel estimation. Closed-form expressions of our proposed schemes are respectively derived and confirmed by means of simulation results, which verify the superiority of B-NOMA, B-RSMA and EB-RSMA compared to the benchmark.