Recently, there has been an increasing focus on audio-text cross-modal learning. However, most of the existing audio-text datasets contain only simple descriptions of sound events. Compared with classification labels, the advantages of such descriptions are significantly limited. In this paper, we first analyze the detailed information that human descriptions of audio may contain beyond sound event labels. Based on the analysis, we propose an automatic pipeline for curating audio-text pairs with rich details1. Leveraging the property that sounds can be mixed and concatenated in the time domain, we control details in four aspects: temporal relationship, loudness, speaker identity, and occurrence number, in simulating audio mixtures. Corresponding details are transformed into captions by large language models. Audio-text pairs with rich details in text descriptions are thereby obtained. We validate the effectiveness of our pipeline with a small amount of simulated data, demonstrating that the simulated data enables models to learn detailed audio captioning.

To facilitate the emerging applications in 5G networks, mobile network operators will provide many network functions in terms of control and prediction. Recently, they have recognized the power of machine learning (ML) and started to explore its potential to facilitate those network functions. Nevertheless, the current ML models for network functions are often derived in an offline manner, which is inefficient due to the excessive overhead for transmitting a huge volume of dataset to remote ML training clouds and failing to provide the incremental learning capability for the continuous model updating. As an alternative solution, we proposeCocktail, an incremental learning framework within a reference 5G network architecture. To achieve cost efficiency while increasing trained model accuracy, an efficient online data scheduling policy is essential. To this end, we formulate an online data scheduling problem to optimize the framework cost while alleviating the data skew issue caused by the capacity heterogeneity of training workers from the long-term perspective. We exploit the stochastic gradient descent to devise an online asymptotically optimal algorithm, including two optimal policies based on novel graph constructions for skew-aware data collection and data training. Small-scale testbed and large-scale simulations validate the superior performance of our proposed framework.

The incoming 5G cellular network is stepping into a densification era, where various kinds of base stations are densely deployed to provide fruitful mobile services such as video streaming. In order to improve the performance of these mobile services, the way to optimally allocate cellular resources for the network-wide users is a crucial problem. In this paper, we consider the mobile Ultra-High-Definition (UHD) video streaming service, envision a 5G ultra dense network (UDN) consisting of a series of Video Base Stations (VBSs) dedicated to video streaming services for multiple users, and mainly explore the impact of cellular resource allocation on video streaming. We incorporate two important video streaming states into cellular resource allocation and formulate a streaming-aware and fairness-aware cellular resource allocation problem. To deal with the formulated problem, we provide a novel graph transformation and design an optimal matching algorithm which can be solved in polynomial time. Extensive trace-driven simulations validate the superior performance of the proposed algorithm under different user mobility patterns and network scales.

Ultra-High-Definition (UHD) videos have absorbed great attention in recent years. However, as they are of significant size, streaming them require an extremely high bandwidth to achieve a good quality of experience, which poses a great challenge on the current cellular networks. Realizing the great potentials of coordinated multi-point joint transmission (JT-CoMP) in 5G Ultra Dense Network, we propose a novel Tuner framework for the UHD video streaming service. In this framework, we strive to design an efficient algorithm for VBS sleeping and VBS grouping to maximize the data rates of overall video users while reducing the overhead of cellular networks. To this end, we provide a comprehensive framework model and formulate a single-timescale VBS sleeping & grouping problem. We design a novel master-slave based algorithm to solve the formulated mixed-integer programming problem optimally with low complexity. In addition, we extend it to facilitate the more practical setting, i.e., two-timescale VBS sleeping & grouping. Extensive simulations validate the superior performance of our framework in various system settings.