Predictive intelligence to the edge: impact on edge analytics

Abstract

We rest on the edge computing paradigm where pushing processing and inference to the edge of the Internet of Things (IoT) allows the complexity of predictive analytics to be distributed into smaller pieces physically located at the source of the contextual information. This enables a huge amount of rich contextual data to be processed in real time that would be prohibitively complex and costly to deliver on a traditional centralized Cloud. We propose a lightweight, distributed, predictive intelligence mechanism that supports communication efficient aggregation and predictive modeling within the edge network. Our idea is based on the capability of the edge nodes to (1) monitor the evolution of the sensed time series contextual data, (2) locally determine (through prediction) whether to disseminate contextual data in the edge network or not, and (3) locally re-construct undelivered contextual data in light of minimizing the required communication interaction at the expense of accurate analytics tasks. Based on this on-line decision making, we eliminate data transfer at the edge of the network, thus saving network resources by exploiting the evolving nature of the captured contextual data. We provide comprehensive analytical, experimental and comparative evaluation of the proposed mechanism with other mechanisms found in the literature over real contextual datasets and show the benefits stemmed from its adoption in edge computing environments.

Internet of Things and data mining: From applications to techniques and systems

“…Harth, Anagnostopoulos, and Pezaros ()) proposed a lightweight and distributed prediction method for efficient data aggregation on the edge. The scheme also enables efficient predictive modeling within the distributed edge computing system.…”

Quality-aware aggregation & predictive analytics at the edge

“…However, such approaches are unaware of communication efficiency in the edge network as supported by the above-mentioned selective forwarding approaches. Our previous work [12] investigates the impact of a predictionbased selective forwarding decision, purely from the communication objective, on aggregation and predictive analytics. This signals the necessity of introducing a hybrid and sophisticated decision making model on when & which data to process and deliver for trading between quality of (advanced) analytics and communication efficiency at the network edge.…”

“…• two real-time model variants exploiting the computational capabilities of the collaborating edge devices over real contextual data streams; • comparative & performance assessment with aggregation and linear regression models using statistical & information theoretic metrics comparing our model with the methodologies [12], [4], [5], [6] following the selective forwarding scheme; The paper is organized as follows: Section II discusses the rationale and provides fundamental definitions for the quality analytics metrics, while Section III presents the overall approach and problem formulation. Section IV elaborates on the solution fundamentals, while Section V reports on the performance and comparative assessment.…”

“…The methodologies in [12], [4], [5], and [6] adopt a selective forwarding rule to decide whether to deliver a context vector in the edge network or not in light of minimizing the communication overhead defying the quality of analytics tasks. Such methodologies are based on an Instantaneous Decision Making (IDM) using only (i) the current vector x t and (ii) the expected (predicted) vectorx t .…”

“…We compare the performance of the OVF and HOVF variants with the models in [12], [4], [5], and [6], which implement the IDM methodology, over real contextual data described in [21]. The dataset contains T = 10 4 context vectors in a 12-dimensional (d = 12) real data space corresponding to sensing air quality parameters reflecting 12 SANs of an edge network connecting with EN.…”