Equilibrium in the Computing Continuum through Active Inference

被引:9
作者
Sedlak, Boris [1 ]
Pujol, Victor Casamayor [1 ]
Donta, Praveen Kumar [1 ]
Dustdar, Schahram [1 ]
机构
[1] TU Wien, Distributed Syst Grp, A-1040 Vienna, Austria
来源
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE | 2024年 / 160卷
关键词
Active Inference; Computing Continuum; Scalability; Edge intelligence; Transfer learning; Equilibrium; CAUSAL;
D O I
10.1016/j.future.2024.05.056
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Computing Continuum (CC) systems are challenged to ensure the intricate requirements of each computational tier. Given the system's scale, the Service Level Objectives (SLOs), which are expressed as these requirements, must be disaggregated into smaller parts that can be decentralized. We present our framework for collaborative edge intelligence, enabling individual edge devices to (1) develop a causal understanding of how to enforce their SLOs and (2) transfer knowledge to speed up the onboarding of heterogeneous devices. Through collaboration, they (3) increase the scope of SLO fulfillment. We implemented the framework and evaluated a use case in which a CC system is responsible for ensuring Quality of Service (QoS) and Quality of Experience (QoE) during video streaming. Our results showed that edge devices required only ten training rounds to ensure four SLOs; furthermore, the underlying causal structures were also rationally explainable. The addition of new types of devices can be done a posteriori; the framework allowed them to reuse existing models, even though the device type had been unknown. Finally, rebalancing the load within a device cluster allowed individual edge devices to recover their SLO compliance after a network failure from 22% to 89%.
引用
收藏
页码:92 / 108
页数:17
相关论文
共 53 条
[41]   Designing Reconfigurable Intelligent Systems with Markov Blankets [J].
Sedlak, Boris ;
Pujol, Victor Casamayor ;
Donta, Praveen Kumar ;
Dustdar, Schahram .
SERVICE-ORIENTED COMPUTING, ICSOC 2023, PT I, 2023, 14419 :42-50
[42]   A Privacy Enforcing Framework for Data Streams on the Edge [J].
Sedlak, Boris ;
Murturi, Ilir ;
Donta, Praveen Kumar ;
Dustdar, Schahram .
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING, 2024, 12 (03) :852-863
[43]   A step-by-step tutorial on active inference and its application to empirical data [J].
Smith, Ryan ;
Friston, Karl J. ;
Whyte, Christopher J. .
JOURNAL OF MATHEMATICAL PSYCHOLOGY, 2022, 107
[44]   Edge2Train: A Framework to Train Machine Learning Models (SVMs) on Resource-Constrained IoT Edge Devices [J].
Sudharsan, Bharath ;
Breslin, John G. ;
Ali, Muhammad Intizar .
PROCEEDINGS OF 10TH INTERNATIONAL CONFERENCE ON THE INTERNET OF THINGS, IOT 2020, 2020,
[45]   The 6G Computing Continuum (6GCC): Meeting the 6G computing challenges [J].
Tarneberg, William ;
Fitzgerald, Emma ;
Bhuyan, Monowar ;
Townend, Paul ;
Arzen, Karl-Erik ;
Ostberg, Per-Olov ;
Elmroth, Erik ;
Eker, Johan ;
Tufvesson, Fredrik ;
Kihl, Maria .
2022 1ST INTERNATIONAL CONFERENCE ON 6G NETWORKING (6GNET), 2022,
[46]  
Tsamardinos I., 2003, TIME SAMPLE EFFICIEN
[47]  
Tschantz A., 2020, Reinforcement Learning through Active Inference
[48]   Updating conditional probabilities of Bayesian belief networks by merging expert knowledge and system monitoring data [J].
Vagnoli, Matteo ;
Remenyte-Prescott, Rasa .
AUTOMATION IN CONSTRUCTION, 2022, 140
[49]   A Novel Algorithm for Merging Bayesian Networks [J].
Vanis, Miroslav ;
Lokaj, Zdenek ;
Srotyr, Martin .
SYMMETRY-BASEL, 2023, 15 (07)
[50]  
Vowels M. J., 2021, D'ya like DAGs? A Survey on Structure Learning and Causal Discovery