Spatio-Temporal Aware Collaborative Service Ranking Prediction in IoT-Enabled Edge Computing
With the rapid proliferation of services deployed on edge servers, selecting high quality services from a multitude of functionally similar offerings based on their rankings has become a critical issue. Service ranking is an effective approach to solve this problem. Traditional QoS aware service ran...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11021644/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | With the rapid proliferation of services deployed on edge servers, selecting high quality services from a multitude of functionally similar offerings based on their rankings has become a critical issue. Service ranking is an effective approach to solve this problem. Traditional QoS aware service ranking approaches encounter new challenges in edge computing enabled Internet of Things (IoT) systems. On one hand, the geographical characteristics of services introduce heterogeneity into the evaluation criteria for service ranking. On the other hand, analyzing the multi-dimensional temporal characteristics of QoS poses a significant challenge. To address these challenges, we propose a spatio-temporal aware collaborative service ranking prediction approach, named ST-CRank, to achieve accurate service ranking results. Specifically, a spatial-aware service partial ranking model is introduced to generate partial rankings. In this model, a silhouette coefficient based clustering algorithm is utilized to partition edge domains, and services deployed on the same edge server are compared using the partial ranking model. The QoS comparisons are forecasted using a deep time series model to obtain partial rankings for each edge server, thereby capturing the multi-dimensional temporal characteristics of QoS. With the partial rankings obtained, the global ranking is achieved by aggregating the partial rankings within the same edge domain. Finally, the effectiveness of ST-CRank is evaluated through large scale real world dataset based experiments. The results demonstrate that our approach achieves higher accuracy in prediction compared to other baseline algorithms. |
|---|---|
| ISSN: | 2169-3536 |