Edge assisted IoT orchestration

Funder

Organizational Unit

Publications

Attention-based model and deep reinforcement learning for distribution of event processing tasks

Publication . Mazayev, Andriy; Al-Tam, Faroq; Correia, Noélia

Event processing is the cornerstone of the dynamic and responsive Internet of Things (IoT). Recent approaches in this area are based on representational state transfer (REST) principles, which allow event processing tasks to be placed at any device that follows the same principles. However, the tasks should be properly distributed among edge devices to ensure fair resources utilization and guarantee seamless execution. This article investigates the use of deep learning to fairly distribute the tasks. An attention-based neural network model is proposed to generate efficient load balancing solutions under different scenarios. The proposed model is based on the Transformer and Pointer Network architectures, and is trained by an advantage actorcritic reinforcement learning algorithm. The model is designed to scale to the number of event processing tasks and the number of edge devices, with no need for hyperparameters re-tuning or even retraining. Extensive experimental results show that the proposed model outperforms conventional heuristics in many key performance indicators. The generic design and the obtained results show that the proposed model can potentially be applied to several other load balancing problem variations, which makes the proposal an attractive option to be used in real-world scenarios due to its scalability and efficiency.

2022-07Journal article

Open Access

Event processing in web of things

Publication . Mazayev, Andriy; Correia, N.

The incoming digital revolution has the potential to drastically improve our productivity, reduce operational costs and improve the quality of the products. However, the realization of these promises requires the convergence of technologies — from edge computing to cloud, artificial intelligence, and the Internet of Things — blurring the lines between the physical and digital worlds. Although these technologies evolved independently over time, they are increasingly becoming intertwined. Their convergence will create an unprecedented level of automation, achieved via massive machine-to-machine interactions whose cornerstone are event processing tasks. This thesis explores the intersection of these technologies by making an in-depth analysis of their role in the life-cycle of event processing tasks, including their creation, placement and execution. First, it surveys currently existing Web standards, Internet drafts, and design patterns that are used in the creation of cloud-based event processing. Then, it investigates the reasons for event processing to start shifting towards the edge, alongside with the standards that are necessary for a smooth transition to occur. Finally, this work proposes the use of deep reinforcement learning methods for the placement and distribution of event processing tasks at the edge. Obtained results show that the proposed neural-based event placement method is capable of obtaining (near) optimal solutions in several scenarios and provide hints about future research directions.

2022-05-30Doctoral thesis

Open Access

A distributed CoRE-Based resource synchronization mechanism

Publication . Mazayev, A.; Correia, Noélia

Representational state transfer (REST) application programming interfaces and event processing are the cornerstone of the dynamic Internet of Things. While the former is required for device interoperability, the latter is important for autonomous and responsive systems. In recent years, both topics have received a lot of attention and have been drastically changing due to the emergence of new applications, which end up working inefficiently with current standards and architectures. More recently, event processing started to move down from the top (cloud) to bottom (edge devices). At the same time, the Internet Engineering Task Force, which normally solves low-layer protocol-related problems, has also started looking at event processing and resource synchronization from a bottom-up perspective. This article explores the intersection of these efforts by making an in-depth overview of currently existing standards, and Internet drafts, that allow building complex event processing chains. Next, a new reusable and scalable event processing mechanism, which can be distributed across multiple end-devices, is introduced. Its optimal distribution across end-devices is mathematically addressed, and results confirm its effectiveness.

2020-05Journal article

Restricted