Distributed DRL-Based Computation Offloading Scheme for Improving QoE in Edge Computing Environments.

Various edge collaboration schemes that rely on reinforcement learning (RL) have been proposed to improve the quality of experience (QoE). Deep RL (DRL) maximizes cumulative rewards through large-scale exploration and exploitation. However, the existing DRL schemes do not consider the temporal states using a fully connected layer. Moreover, they learn the offloading policy regardless of the importance of experience. They also do not learn enough because of their limited experiences in distributed environments. To solve these problems, we proposed a distributed DRL-based computation offloading scheme for improving the QoE in edge computing environments. The proposed scheme selects the offloading target by modeling the task service time and load balance. We implemented three methods to improve the learning performance. Firstly, the DRL scheme used the least absolute shrinkage and selection operator (LASSO) regression and attention layer to consider the temporal states. Secondly, we learned the optimal policy based on the importance of experience using the TD error and loss of the critic network. Finally, we adaptively shared the experience between agents, based on the strategy gradient, to solve the data sparsity problem. The simulation results showed that the proposed scheme achieved lower variation and higher rewards than the existing schemes.

Reinforcement Learning-Based Adaptive Streaming Scheme with Edge Computing Assistance.

Dynamic Adaptive Streaming over HTTP (DASH) is a promising scheme for improving the Quality of Experience (QoE) of users in video streaming. However, the existing schemes do not perform coordination among clients and depend on fixed heuristics. In this paper, we propose an adaptive streaming scheme with reinforcement learning in edge computing environments. The proposed scheme improves the overall QoE of clients and QoE fairness among clients based on a state-of-the-art reinforcement learning algorithm. Edge computing assistance plays a role in providing client-side observations to the mobile edge, making agents utilize this information when generating a policy for multi-client adaptive streaming. We evaluated the proposed scheme through simulation-based experiments under various network conditions. The experimental results show that the proposed scheme achieves better performance than the existing schemes.

Robust Data Augmentation Generative Adversarial Network for Object Detection.

Generative adversarial network (GAN)-based data augmentation is used to enhance the performance of object detection models. It comprises two stages: training the GAN generator to learn the distribution of a small target dataset, and sampling data from the trained generator to enhance model performance. In this paper, we propose a pipelined model, called robust data augmentation GAN (RDAGAN), that aims to augment small datasets used for object detection. First, clean images and a small datasets containing images from various domains are input into the RDAGAN, which then generates images that are similar to those in the input dataset. Thereafter, it divides the image generation task into two networks: an object generation network and image translation network. The object generation network generates images of the objects located within the bounding boxes of the input dataset and the image translation network merges these images with clean images. A quantitative experiment confirmed that the generated images improve the YOLOv5 model's fire detection performance. A comparative evaluation showed that RDAGAN can maintain the background information of input images and localize the object generation location. Moreover, ablation studies demonstrated that all components and objects included in the RDAGAN play pivotal roles.

Preference-Tree-Based Real-Time Recommendation System.

In the current era of online information overload, recommendation systems are very useful for helping users locate content that may be of interest to them. A personalized recommendation system presents content based on information such as a user's browsing history and the videos watched. However, information filtering-based recommendation systems are vulnerable to data sparsity and cold-start problems. Additionally, existing recommendation systems suffer from the large overhead incurred in learning regression models used for preference prediction or in selecting groups of similar users. In this study, we propose a preference-tree-based real-time recommendation system that uses various tree models to predict user preferences with a fast runtime. The proposed system predicts preferences based on two balance constants and one similarity threshold to recommend content with a high accuracy while balancing generalized and personalized preferences. The results of comparative experiments and ablation studies confirm that the proposed system can accurately recommend content to users. Specifically, we confirmed that the accuracy and novelty of the recommended content were, respectively, improved by 12.1% and 27.2% compared to existing systems. Furthermore, we verified that the proposed system satisfies real-time requirements and mitigates both cold-start and overfitting problems.

Domain-Specific On-Device Object Detection Method.

Object detection is a significant activity in computer vision, and various approaches have been proposed to detect varied objects using deep neural networks (DNNs). However, because DNNs are computation-intensive, it is difficult to apply them to resource-constrained devices. Here, we propose an on-device object detection method using domain-specific models. In the proposed method, we define object of interest (OOI) groups that contain objects with a high frequency of appearance in specific domains. Compared with the existing DNN model, the layers of the domain-specific models are shallower and narrower, reducing the number of trainable parameters; thus, speeding up the object detection. To ensure a lightweight network design, we combine various network structures to obtain the best-performing lightweight detection model. The experimental results reveal that the size of the proposed lightweight model is 21.7 MB, which is 91.35% and 36.98% smaller than those of YOLOv3-SPP and Tiny-YOLO, respectively. The f-measure achieved on the MS COCO 2017 dataset were 18.3%, 11.9% and 20.3% higher than those of YOLOv3-SPP, Tiny-YOLO and YOLO-Nano, respectively. The results demonstrated that the lightweight model achieved higher efficiency and better performance on non-GPU devices, such as mobile devices and embedded boards, than conventional models.

Resource Prediction-Based Edge Collaboration Scheme for Improving QoE.

Recent years have witnessed a growth in the Internet of Things (IoT) applications and devices; however, these devices are unable to meet the increased computational resource needs of the applications they host. Edge servers can provide sufficient computing resources. However, when the number of connected devices is large, the task processing efficiency decreases due to limited computing resources. Therefore, an edge collaboration scheme that utilizes other computing nodes to increase the efficiency of task processing and improve the quality of experience (QoE) was proposed. However, existing edge server collaboration schemes have low QoE because they do not consider other edge servers' computing resources or communication time. In this paper, we propose a resource prediction-based edge collaboration scheme for improving QoE. We estimate computing resource usage based on the tasks received from the devices. According to the predicted computing resources, the edge server probabilistically collaborates with other edge servers. The proposed scheme is based on the delay model, and uses the greedy algorithm. It allocates computing resources to the task considering the computation and buffering time. Experimental results show that the proposed scheme achieves a high QoE compared with existing schemes because of the high success rate and low completion time.

An efficient transmission power control scheme for temperature variation in wireless sensor networks.

Wireless sensor networks collect data from several nodes dispersed at remote sites. Sensor nodes can be installed in harsh environments such as deserts, cities, and indoors, where the link quality changes considerably over time. Particularly, changes in transmission power may be caused by temperature, humidity, and other factors. In order to compensate for link quality changes, existing schemes detect the link quality changes between nodes and control transmission power through a series of feedback processes, but these approaches can cause heavy overhead with the additional control packets needed. In this paper, the change of the link quality according to temperature is examined through empirical experimentation. A new power control scheme combining both temperature-aware link quality compensation and a closed-loop feedback process to adapt to link quality changes is proposed. We prove that the proposed scheme effectively adapts the transmission power to the changing link quality with less control overhead and energy consumption.