Special Issue on Advanced Wireless Sensor Networks for Emerging Applications

1. Introduction

Wireless sensor networks (WSNs) have been widely used due to their extensive range of applications. The main role of WSNs can be accounted for by monitoring and collecting relevant data with distributed sensor nodes in the coverage area. Therefore, typical studies on WSNs have dealt with network configuration, network operation, and performance evaluation of WSNs executing the main role. Thanks to the continual evolution of sensor technology enabling the implementation of small, compact, and computationally efficient sensor devices, internet of things (IoT) connectivity becomes a reality. While monitoring and collecting sensor data alone do not involve significant local data processing for decision making at sensor nodes, some emerging applications of WSNs require fast and reliable on-device decision making. Due to massively deployed IoT devices and some of them demanding large bandwidth incurring issues with net-working connectivity, on-device decision making inevitably introduces edge computing capability of sensor nodes for efficient local data processing [1]. The benefit of edge computing increases when the volume of collected sensor data for subsequent transmission to a data center becomes excessively large. The recent development of deep learning techniques contributes to the enhancement of edge computing capability, and as a result applications of WSNs with edge computing capability, they have become broader than ever. This editorial reviews emerging applications of WSNs and their key technical components and two contributions to this Special Issue.

2. Advanced Wireless Sensor Networks for Emerging Applications

The WSNs are widely deployed in many technical areas. For smart cities, the WSNs can be applied with the IoT technologies to collect diverse information from different locations, improving the quality of life of people. The infrastructure WSNs, where the sensor nodes and sink node(s) are static, can be utilized for structural health monitoring of buildings, environmental monitoring in greenhouses, monitoring of water resource, and smart metering of utilities for homes and buildings. In [2], the deployment of the WSN is presented for efficient and effective structural health monitoring. In the field of agriculture, the WSN is used to measure soil moisture, light, humidity, and temperature [3]. The online water contaminant monitoring system preventing contaminant intrusion is implemented with the WSN [4]. In [5], the performance of in-home energy management (iHEM) based on the WSN for the home area is evaluated, allowing communication between a user and a controller for the convenience of the user. The iHEM helps reduce the electricity cost, peak load, and carbon emissions. The mobile WSNs have at least one mobile sensor node. A vehicle can be taken as a mobile sensor for traffic monitoring and can become a sensor node of the vehicular ad hoc sensor network [6]. In [7], the system for monitoring road and traffic conditions has been designed, using the data from the global positioning system (GPS) and accelerometer of the smartphone (sensor device). To establish more advanced WSNs with mobile sensor nodes, challenging tasks, such as the routing planning problem of mobile nodes, data routing problem of sensor nodes, and task assignment problem of heterogeneous nodes, should be considered.

The range of applications of the WSNs includes small-scale power grids. For microgrids providing functionality to control and monitor distribution power systems, the WSNs can be used for a broad range of options, such as wireless meter reading, fault diagnostics and self-maintenance of equipment, and remote monitoring of distribution power system. In [8], a residential parking station is integrated to the microgrid, and the charging/discharging operation of electric vehicles are determined based on the measured data with the WSN, such as local load demand and photovoltaic (PV) power production. In [9], an energy-efficient coverage-preserving scheduling algorithm is proposed to extend the lifetime of the WSN by scheduling the activation of each set of sensor nodes. In [10], the WSNs are engaged to monitor the status of power generation, transmission, and distribution, and a fog computing network is established for the optimization of WSN in a microgrid to reduce energy consumption and improve the packet delivery ratio as well as the throughput. For the improved operation of a nanogrid, e.g., local power grid in kWatt scale, data transmission through the smart meter, real-time control, accurate decision making, and energy management can be achieved based on the advanced communication technologies in WSN [11].

Smart charging schemes are crucial for proper operation of rechargeable WSNs. In fact, techniques for charging batteries have a range of applications including supply of electric power to railway vehicles [12]. Due to the limited lifetime of batteries of the WSNs, periodic or aperiodic recharging of sensor nodes along a predetermined path is required. Optimal path planning, which is also important for minimized energy consumption of vehicle operation [13], is critical to save time for charging entire sensor nodes. Various charging schemes [14,15,16] have been reported in the literature. An energy efficient geometric routing (GR) protocol is presented to reduce the power consumption and time by calculating a traversing path of wireless charging vehicle [17]. The geometric solution, applied to the GR protocol, divides the network topology into a grid structure and uses a one-to-many recharging scheme. To find the shortest path of the mobile charger, which can fully recharge all sensor nodes by passing them only once, a greedy constructing tree algorithm (GCTA) is proposed in [18]. The GCTA constructs the shortest path with low complexity by using the Markov chain theorem, graph theory, and tree theory.

The intelligent transportation system has evolved into a tool for supplying traffic information as well as traffic control, necessitating the use of contemporary and effective measurement methodologies. The road conditions, collected by static sensor nodes on the road or previously passed vehicles, can be delivered to all vehicles which access them through the vehicular sensor network [19]. A traffic surveillance and speed violation detection system can be established with high precision by deploying the collaborative WSNs, instead of using a traffic enforcement vehicle [20]. Additionally, the optimal path for a specific destination can be extracted based on the monitoring of small- to medium-sized road networks. In [21], the WSN is deployed to the smart parking system to provide the detection of vehicles, reservation of parking spaces, and guidance to external systems. In an autonomous vehicle, the light detection and ranging (LiDAR) sensor plays a critical role because it can provide the perception of all light circumstances. The vehicle-to-vehicle communication, LiDAR system, and GPS are used to construct sensor fusion algorithms in connected-vehicles applications for more accurate localization and object detection. In order to extract dense features efficiently in the large raw data sets collected by LiDARs, a novel framework for point cloud registration is developed in [22], which uses the graph attention network as an attention mechanism enriching the relationships between point clouds. The data can be interchanged between roadside and vehicles or between vehicles to prevent car accident and control road network, which can be translated as the mechanism of the WSN.

Machine learning (ML) techniques are being applied for more advanced WSNs [23]. The use of ML techniques is beneficial for enhanced edge computing at sensor devices. In the IoT systems, the ML technologies are mainly used for node localization, coverage and connectivity problems, medium access control (MAC) and routing protocols, data collection, fault detection, energy harvesting, and monitoring system. For node localization, which is significant for the path planning of mobile sensor nodes, some ML algorithms, such as the support vector machine, k-nearest neighbor algorithm, reinforcement learning (RL)-based algorithms [24,25], are applied to identify the current location of a sensor node [26]. The RL algorithm particularly corresponds to the deep learning (DL) algorithm. In [27], a multi-hop ad hoc IoT network is constructed to determine the transmission ranges of wireless sensors based on the RL approach for appropriate connectivity between sensors and maximization of the lifetime of the WSN. A sensor device is used as a decision-making agent deciding the transmission range, which maximizes the network throughput and minimizes the transmission energy consumption. An energy efficient routing protocol based on the RL is proposed to maintain a stable WSN by distributing residual energy of sensor nodes more evenly [28]. The reward of the RL, utilized to select appropriate forwarders of packets, is calculated based on the residual energy level of each sensor node as well as the energy distribution in a group of sensor nodes. In [29], a combined prediction model, which combines a back propagation neural network with a genetic algorithm and particle swarm optimization, respectively, is used for wind farm energy harvesting based on the weather data gathered by the WSN. Additionally, for efficient data transmission of PV and wind power production data predicted by kernel recursive least-square (KRLS) algorithm, a link scheduling method based on sensor node attributes is presented in [30].

3. Conclusions

In order to maintain an energy efficient and sustainable WSN on a large scale, technical advancement relevant to the design, implementation, and deployment of WSNs is essential. It is found that advanced techniques developed for WSNs improve the networking performance, energy efficiency, and operation stability of WSNs in many emerging applications. Especially, ML or DL techniques are highly utilized for advanced WSNs capable of edge computing. It is expected that advanced approaches for WSN applications will inspire future studies on diverse multi-disciplinary research areas.