Publications

• Ruslan Kain, Yuanzhu Chen, and Hossam S. Hassanein. 2024. "Predictive Resource Usage Characterization for Extreme Edge Computing" IEEE Canadian Conference on Electrical and Computer Engineering (CCECE), 06 August, 2024, Pages 93-94.
• Ruslan Kain, Sara A. Elsayed, Yuanzhu Chen, and Hossam S. Hassanein. 2023. "DRUDGE: Dynamic Resource Usage Data Generation for Extreme Edge Devices" IEEE Global Communications Conference (GLOBECOM), 04-08 December, 2023.
• Ruslan Kain, Sara A. Elsayed, Yuanzhu Chen, and Hossam S. Hassanein. 2023. "RUMP: Resource Usage Multi-Step Prediction in Extreme Edge Computing" Computer Communications, Volume 210, 1 October 2023, Pages 45-57.
• Ruslan Kain, Sara A. Elsayed, Yuanzhu Chen, and Hossam S. Hassanein. 2022. "Multi-step Prediction of Worker Resource Usage at the Extreme Edge" Association for Computing Machinery (ACM), In Proceedings of the 25th International ACM Conference on Modeling Analysis and Simulation of Wireless and Mobile Systems (MSWiM '22), 25–32.
• Ruslan Kain and Sameh Sorour. 2022."Worker Resource Characterization Under Dynamic Usage in Multi-access Edge Computing" IEEE, 2022 International Wireless Communications and Mobile Computing (IWCMC), 1070-1075
• Ruslan Kain and Hazem Hajj. 2021."An Optimization Approach to Multi-Sensor Operation for Multi-Context Recognition"" Sensors 2021, 21, 6862.
• Ruslan Kain, Daniele Manerba, and Roberto Tadei. 2021"The index selection problem with configurations and memory limitation: A scatter search approach" Computers & Operations Research, v133, 105385.
• Gilbert Badaro, Obeida El Jundi, Alaa Khaddaj, Alaa Maarouf, Raslan Kain, Hazem Hajj, and Wassim El-Hajj. 2018. "EMA at SemEval-2018 task 1: Emotion mining for Arabic" Association for Computational Linguistics (ACL), In Proceedings of the 12th International Workshop on Semantic Evaluation, 236–244

Data

The dataset includes dynamic resource usage information associated with running edge-native applications on a set of four heterogeneous Raspberry Pi 4 Devices. The four Raspberry Pi 4 devices have 2, 4, and 8 GB RAM sizes, and CPU frequencies of 1200, 1500, and 1800 MHz. The resource usage measurements have a five-second granularity. We managed to collect more than 550 thousand unique data points representing the 768 hours of running applications on Raspberry Pi Devices. Dataset 1

Multi-Sensor Operation information for Multi-Context Recognition Scenario: Borealis Dataset

Codebase

Contains script for resource usage data generation on dynamically accessed Raspberry Pis used in DRUDGE. Analysis of the data and multi-step multi-variate prediction methods used in RUMP: rump-ec

Experimental Setups

DRUDGE, a Python-based tool, is designed to simulate the dynamic usage behavior of Extreme Edge Devices (EEDs), also known as workers, in a controlled, repeatable manner. This tool supports a wide variety of applications and allows for the customization of application run times, CPU frequencies, and network conditions to generate diverse datasets covering various usage scenarios. It employs Raspberry Pi (RPi) 4B model devices with different RAM sizes and CPU frequencies, modified through overclocking and throttling to increase heterogeneity. Applications such as video gaming (Doom), video streaming (YouTube), augmented reality simulations, and Duino-coin mining are run on these devices to emulate dynamic resource usage. DRUDGE enables users to input parameters like monitoring interval, CPU frequency, and dataset name/type, creating state time lengths for application runs based on these inputs. Tailored specifically for Extreme Edge Computing (EEC) applications, the tool collects data over multiple 48-hour periods with 5-second monitoring intervals, recording resource usage with the psutil Python library and adapting to different network conditions to reflect real-world IoT scenarios.