Figure 2 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Source publication
Wireless networks are trending towards large scale systems, containing thousands of nodes, with multiple co-existing applications. Congestion is an inevitable consequence of this scale and complexity, which leads to inefficient use of the network capacity. This paper proposes an autonomous and adaptive wireless network management framework, utilisi...
Context in source publication
Context 1
... timely change in the environment, a complete state space, and a suitable reward function expedite the policy update or learning process. Reward r t+1 Compared with single-agent systems, multi-agent systems differ in that there are multiple agents interacting with a common environment, as illustrated in Figure 2. In multiagent systems, the agents can all share the same perspective of the environment, where they all receive the same state information, or they can each have a distinct, partial view of the environment. ...
Similar publications
Lane-change maneuver has always been a challenging task for both manual and autonomous driving, especially in an urban setting. In particular, the uncertainty in predicting the behavior of other vehicles on the road leads to indecisive actions while changing lanes, which, might result in traffic congestion and cause safety concerns. This paper anal...
Citations
... This algorithm divided the physical resources of the gateway into multiple virtual networks to provide different QoS guarantee. The authors in [33] provided an approach to the network management under multiple co-existing applications using MARL. Their results revealed that MARL can significantly improve ADR and the responsiveness compared with the single-agent model. ...
... There are mainly three types of approaches. (II-a) Deep reinforcement learning is a machine-learning algorithm that deals with a type of problem where a learner performs a certain state and receives a certain reward [202][203][204][205][206][207][208][209][210][211][212][213]. By repeatedly selecting actions and receiving rewards, it learns decision-making strategies for selecting actions that yield more rewards in the future. ...
The adoption of “artificial intelligence (AI) in drug discovery”, where AI is used in the process of pharmaceutical research and development, is progressing. By using the ability to process large amounts of data, which is a characteristic of AI, and achieving advanced data analysis and inference, there are benefits such as shortening development time, reducing costs, and reducing the workload of researchers. There are various problems in drug development, but the following two issues are particularly problematic: (1) the yearly increases in development time and cost of drugs and (2) the difficulty in finding highly accurate target genes. Therefore, screening and simulation using AI are expected. Researchers have high demands for data collection and the utilization of infrastructure for AI analysis. In the field of drug discovery, for example, interest in data use increases with the amount of chemical or biological data available. The application of AI in drug discovery is becoming more active due to improvement in computer processing power and the development and spread of machine-learning frameworks, including deep learning. To evaluate performance, various statistical indices have been introduced. However, the factors affected in performance have not been revealed completely. In this study, we summarized and reviewed the applications of deep learning for drug discovery with BigData.
LoRaWAN is a low-power wide-area network standard widely used for long-range machine-to-machine communications in the Internet of Things ecosystem. While enabling ultra-low-power communication links, its open nature impulsed exponential market growth in the last years. Given its Aloha-like medium access nature, several scalability-oriented improvements were proposed in the last years, with time-slotted communications having raised special interest. However, how to efficiently allocate resources in a network where the cost of downlink communication is significantly higher than that of the uplink represents a significant challenge. To shed light on this matter, this work proposes the use of multi-agent systems for network planning in time-slotted communications. To do so, a predictive network planning agent is designed and validated as part of an end-to-end multi-agent network management system, which is based on multi-output regression that predicts the resulting network scalability for a given set of joining devices and setup scenarios being considered. A preliminary evaluation of network-status predictions showed a mean absolute error lower than 3% and pointed out different lessons learned, in turn validating the feasibility of the proposed agent for LoRaWAN-oriented network planning.KeywordsMASLoRaWANNetwork PredictionScheduling
Internet of Everything (IoE) is playing an increasingly indispensable role in modern intelligent applications. These smart applications are known for their real-time requirements under limited network and computing resources, in which it becomes a high consuming task to transform and compute tremendous amount of raw data in cloud center. The edge-cloud computing infrastructure allows large amount of data to be processed on nearby edge nodes and then only the extracted and encrypted key features are transmitted to the data center. This offers the potential to achieve an edge-cloud based big data intelligence for IoE in a typical two-stage data processing scheme, while satisfying data security constraint. In this study, a deep reinforcement learning enhanced scheduling method is proposed to address the NP-hard challenge of two-stage scheduling, which is able to allocate computing resources within an edge-cloud infrastructure to ensure computing task to be completed with minimum cost. The proposed reinforcement learning algorithm, which incorporates the Johnson’s rule, is designed to achieve an optimal schedule in IoE. The performance of our method is evaluated and compared with several existing scheduling techniques, and experiment results demonstrate the ability of our proposed algorithm in achieving a more efficient schedule with 1.1-approximation to the targeted optimal IoE applications.
