With the rapid development of ultra-dense networks (UDNs) and random access technologies, device-to-device (D2D) and non-orthogonal multiple access (NOMA) techniques will incorporated into future UDNs supporting both human-type communications (HTC) and machine-type communications (MTC) to fulfill the stringent requirements brought by various potential Internet of Everything (IoE) applications. Nevertheless, the combination of D2D and NOMA will make the network management more complicated. In view of this, we optimize the received signal strength (RSS) threshold value of each small base stations (SBSs) in the UDN where HTC and MTC coexist. Considering the computational complexity, we employ a multi-agent reinforcement learning based RSS threshold value selection scheme, in which each SBS acts as an agent and chose the optimal RSS threshold value to achieve maximum system throughput performance by interacting with the environment. Extensive numerical results show our proposed scheme can greatly improve the system throughput by enhancing the connectivity of massive HTC users and MTC devices via D2D and NOMA techniques.

Visual contents, including images and videos, are dominant on the Internet today. The conventional search engine is mainly designed for textual documents, which must be extended to process and manage increasingly high volumes of visual data objects. In this paper, we present Mixer, an effective system to identify and analyze visual contents and to extract their features for data retrievals, aiming at addressing two critical issues: (1) efficiently and timely understanding visual contents, (2) retrieving them at high precision and recall rates without impairing the performance. In Mixer, the visual objects are categorized into different classes, each of which has representative visual features. Subsystems for model production and model execution are developed. Two retrieval layers are designed and implemented for images and videos, respectively. In this way, we are able to perform aggregation retrievals of the two types in efficient ways. The experiments with Baidu's production workloads and systems show that Mixer halves the model production time and raises the feature production throughput by 9.14x. Mixer also achieves the precision and recall of video retrievals at 95% and 97%, respectively. Mixer has been in its daily operations, which makes the search engine highly scalable for visual contents at a low cost. Having observed productivity improvement of upper-level applications in the search engine, we believe our system framework would generally benefit other data processing applications.