An SVM-Based Method for Classification of External Interference in Industrial Wireless Sensor and Actuator Networks

Abstract

In recent years, the adoption of industrial wireless sensor and actuator networks (IWSANs) has greatly increased. However, the time-critical performance of IWSANs is considerably affected by external sources of interference. In particular, when an IEEE 802.11 network is coexisting in the same environment, a significant drop in communication reliability is observed. This, in turn, represents one of the main challenges for a wide-scale adoption of IWSAN. Interference classification through spectrum sensing is a possible step towards interference mitigation, but the long sampling window required by many of the approaches in the literature undermines their run-time applicability in time-slotted channel hopping (TSCH)-based IWSAN. Aiming at minimizing both the sensing time and the memory footprint of the collected samples, a centralized interference classifier based on support vector machines (SVMs) is introduced in this article. The proposed mechanism, tested with sample traces collected in industrial scenarios, enables the classification of interference from IEEE 802.11 networks and microwave ovens, while ensuring high classification accuracy with a sensing duration below 300 ms. In addition, the obtained results show that the fast classification together with a contained sampling frequency ensure the suitability of the method for TSCH-based IWSAN.

Wi-Fi Meets ML: A Survey on Improving IEEE 802.11 Performance with Machine Learning

“…Among the most important network performance research problems for sensor networks, which can be solved with ML methods, are: sensor grouping (clustering, data aggregation), energyefficient operation (scheduling, duty cycling), resource allocation (cell/channel selection, channel access), traffic classification, routing, mobility prediction, power allocation, interference management, and resource discovery [261]. However, WiFi is only one of many IoT-enabling technologies and 802.11related solutions are rarely mentioned in these surveys with the only directly performance-related work being classifying 802.11 interference using a deep convolutional neural network (DCNN) [264], [265], SVM [266], or various types of SL classifiers: classification trees (CTs) and SVM [267].…”

Smart Campus: An Experimental Performance Comparison of Collaborative and Cooperative Schemes for Wireless Sensor Network

“…An approach sensing the spectrum to identify interference in IIoT devices is [17], but in practical use it takes a large amount of time to identify the issue and provide a solution. A more recent method [18] uses support vector machines that can sense under 300 ms and classifies external interference. In other words, a management system is required to minimize the amount of collisions and increase the effectiveness, such as [19] that uses a self-learning system based on reinforcement learning.…”

Special Issue: Quality of Service in Wireless Sensor/Actuator Networks and Systems

“…Papers cover a wide range of topics, namely the optimization of retransmission scheduling in IEEE 802.15.4e WSANs [1], an experimental evaluation of LoRa reliability [2], the estimation of WSAN lifetime based on innovative battery models [3], a novel radio interference classification method for WSANs [4], a dynamic QoS-aware MAC that can be boosted for long-range communications [5], an RSSI-based model-learning for target localization/tracking [6], using sensor network calculus for designing WSANs with predictable e2e delays [7], and decision-centric WSAN resource management [8]. A brief summary of each paper is provided here.…”

“…While some WSAN applications may cope with lost packets and extra communication delays, when we look into industrial WSAN application contexts, things may radically change, as reliability and timeliness are usually at stake. In this context, the authors of [4] describe a novel method for classifying interference sources in IEEE 802.15.4-based IWSANs, which may then be complemented with interference mitigation techniques. This scheme builds on a machine learning technique (support vector machines) for classifying interference from IEEE 802.11 networks and microwave ovens, as well as the presence of interference-free channels.…”