We consider fault detection in a hydraulic system that maintains multivariate time-series sensor data. Such a real-world industrial environment could suffer from noisy data resulting from inaccuracies in hardware sensing or external interference. Thus, we propose a real-time and robust fault detection method for hydraulic systems that leverages cooperation between cloud and edge servers. The cloud server employs a new approach that includes a genetic algorithm (GA)-based feature selection that identifies feature-to-label correlations and feature-to-feature redundancies. A GA can efficiently process large search spaces, such as solving a combinatorial optimization problem to identify the optimal feature subset. By using fewer important features that require transmission and processing, this approach reduces detection time and improves model performance. We propose a long short-term memory autoencoder for a robust fault detection model that leverages temporal information on time-series sensor data and effectively handles noisy data. This detection model is then deployed at edge servers that provide computing resources near the data source to reduce latency. Our experimental results suggest that this method outperforms prior approaches by demonstrating lower detection times, higher accuracy, and increased robustness to noisy data. While we have a 63% reduction of features, our model obtains a high accuracy of approximately 98% and is robust to noisy data with a signal-to-noise ratio near 0 dB. Our method also performs at an average detection time of only 9.42 ms with a reduced average packet size of 179.98 KB from the maximum of 343.78 KB.
Traffic patterns generated by industrial Internet of Things (IIoT) services can be categorized as either periodic or bursty. The minimal scheduling function (MSF), standardized by the 6TiSCH working group, serves as an example of a scheduling function for IEEE 802.15.4e time-slotted channel hopping. However, the MSF is inadequate for bursty traffic patterns in which a large amount of data is delivered at random intervals. This limitation arises because the MSF requires more dedicated cells to prevent packet loss, but an increased allocation of dedicated cells leads to excessive cell utilization and energy inefficiency. Furthermore, low latency should be considered in bursty traffic patterns, which require proper cell allocation instead of random cell allocation. To address these challenges, we propose a low-latency and Q-learning-based scheduling function (LLQL-SF), is designed for 6TiSCH networks. This new scheduler has been designed to effectively adapt to dynamic traffic patterns by optimizing cell allocation to minimize latency. Additionally, we have integrated a Q-learning algorithm into this scheduler, enabling it to dynamically determine the ideal quantity of dedicated cells required for each slot frame iteration based on the network demands. The proposed methods were evaluated by simulation over various periodic or bursty traffic and network sizes. Also, test over 30 real testbed devices that deployed on FIT IoT-LAB. The results indicated that LLQL-SF outperformed the benchmark methods. LLQL-SF can achieve higher packet delivery, lower latency, and energy usage compared with the standard scheduling function by 11%, -20%, and -11%, respectively.
A smart city is an urban area that collects data from various devices to effectively manage urban resources. The smart city IoT infrastructure connects numerous devices to an Internet-protocol-based low-power wireless network, shares massive amounts of data, and facilitates the development of new services. Message queuing telemetry transport (MQTT), a lightweight exchange protocol for the IoT environment, uses a publish and subscribe structure via a centralized broker to share data. The extent of edge computing provides distributed and closer resources to the data source while maintaining low transmission costs. However, a centralized MQTT data broker is unsuitable for distributed edge resources and could result in high latency, traffic, and bottleneck risk. Therefore, we proposed a distributed MQTT broker optimized architecture. A distributed MQTT broker for edge resources could reduce network traffic and data delivery latency by only managing consumed topics in the network. We formulate an integer non-linear program to optimize container placement and avoid wasting edge computing resources. We compared our proposed architecture to the existing distributed MQTT middleware architecture with greedy and random container placement through extensive simulation. Our methods show better performance in lowering deployment failure ratio, power consumption, network usage, and synchronization overhead.
The 6TiSCH (IPv6 over IEEE802.15.4e time-slotted channel hopping mode) wireless sensor network architecture utilizes control packets to construct network formation. These control packets are essential for establishing communication links between nodes and configuring network settings. The trickle timer algorithm is utilized to broadcast the DIO control packet. DIO carries information about the available parent nodes, which is then used to form the routing tree. Sensors transmit control packets in one cell on each TSCH slotframe, called the minimal cell. This leads to the problem that RPL trickle timer algorithm encounters congestion in DIO control packet transmission with other control messages, particularly in dense networks. Moreover, high traffic transmission also leads to high queue usage, which then drops the DIO control packet. Failed DIO transmission can increase network formation time and energy consumption. To address this issue, we propose Q-Trickle, an adaptive trickle timer algorithm based on Q-learning that determines the optimal policy for transmitting or suppressing DIO based on minimal cell and transmission queue conditions. Q-Trickle adaptively selects a redundancy constant value and transmission interval that promotes fair transmission distribution and considers network condition. Additionally, a control scheme over minimal cell transmission is formulated to lower transmission congestion and faster synchronization. The proposed methods were assessed using simulation and actual testing on the FIT IoT-LAB testbed. The results indicated that Q-Trickle performed better than the benchmark methods. Q-Trickle decreases joining time, energy consumption, and number of failed DIO compared with the original algorithm by -13%, -11%, and -43%, respectively.
The 6TiSCH (IPv6 over the TSCH mode of IEEE 802.15.4e) architecture for wireless sensor networks merges the time-slotted channel hopping (TSCH) at the medium access control (MAC) layer with the routing protocols tailored for low-power and lossy networks (RPL). However, research often neglects the incorporation between TSCH MAC and RPL. Standard RPL strategies rely on an objective function (OF) using the expected transmission count (ETX) metric, which does not adequately reflect the traffic dynamics. Moreover, RPL’s hysteresis function employs a static threshold to control parent change decisions. This static setting disregarded the diverse traffic patterns within the network, leading to unnecessary parent node changes and preventing the node from selecting a better parent. To overcome these shortcomings, we introduce 3 advancements to standard RPL. First, an adaptive parent-changing mechanism based on cooperative Q-learning. Second, a cell usage and traffic load aware objective function. Third, an improved initial transmission cell allocation. Those methods are collectively termed ACI-RPL. We evaluated the performance of the proposed method through simulations using the 6TiSCH simulator and real-hardware tests on the FIT IoT-Lab testbed with OpenWSN firmware. The experiment result indicates that ACI-RPL performs better than the benchmark algorithms. In comparison to the standard RPL, ACI-RPL improves the packet delivery ratio and the total received packets by 12% and 17%, respectively. Additionally, ACI-RPL reduces energy consumption and latency by 23% and 9%.