Edge-assisted Parallel Uncertain Skyline Processing for Low-latency IoE Analysis

TL;DR

This paper introduces EPUS, an edge-assisted parallel algorithm for low-latency uncertain skyline processing in IoE, reducing data transfer and processing time significantly.

Contribution

The paper proposes a novel edge-assisted parallel uncertain skyline algorithm that improves processing latency and reduces data transfer in IoE analytics.

Findings

Reduces latency by over 50% for two-dimensional data.

Outperforms existing methods for high-dimensional data.

Effectively decreases internet data transmission in IoE applications.

Abstract

Due to the Internet of Everything (IoE), data generated in our life become larger. As a result, we need more effort to analyze the data and extract valuable information. In the cloud computing environment, all data analysis is done in the cloud, and the client only needs less computing power to handle some simple tasks. However, with the rapid increase in data volume, sending all data to the cloud via the Internet has become more expensive. The required cloud computing resources have also become larger. To solve this problem, edge computing is proposed. Edge is granted with more computation power to process data before sending it to the cloud. Therefore, the data transmitted over the Internet and the computing resources required by the cloud can be effectively reduced. In this work, we proposed an Edge-assisted Parallel Uncertain Skyline (EPUS) algorithm for emerging low-latency IoE analytic applications. We use the concept of skyline candidate set to prune data that are less likely to become the skyline data on the parallel edge computing nodes. With the candidate skyline set, each edge computing node only sends the information required to the server for updating the global skyline, which reduces the amount of data that transfer over the internet. According to the simulation results, the proposed method is better than two comparative methods, which reduces the latency of processing two-dimensional data by more than 50%. For high-dimensional data, the proposed EPUS method also outperforms the other existing methods.