Multiple wireless communication systems are often operated together in the same area in such manufacturing sites as factories where wideband noise may be emitted from industrial equipment over channels for wireless communication systems. To perform highly reliable wireless communication in such environments, radio wave environments must be monitored that are specific to each manufacturing site to find channels and timing that enable stable communication. The authors studied technologies using machine learning to efficiently analyze a large amount of monitoring data, including signals whose spectrum shape is undefined, such as electromagnetic noise over a wideband. In this paper, we generated common supervised data for multiple sensors by conjointly clustering features after normalizing those calculated in each sensor to recognize the signal reception timing from identical sources and eliminate the complexity of supervised data management. We confirmed our method's effectiveness through signal models and actual data sampled by sensors that we developed.

This paper shows accuracy of using azimuth-variable path-loss fitting in white-space (WS) boundary-estimation. We perform experiments to evaluate this method, and demonstrate that the required number of sensors can be significantly reduced. We have proposed a WS boundary-estimation framework that utilizes sensors to not only obtain spectrum sensing data, but also to estimate the boundaries of the incumbent radio system (IRS) coverage. The framework utilizes the transmitter position information and pathloss fitting. The pathloss fitting describes the IRS coverage by approximating the well-known pathloss prediction formula from the received signal power data, which is measured using sensors, and sensor-transmitter separation distances. To enhance its accuracy, we have further proposed a pathloss-fitting method that employs azimuth variables to reflect the azimuth dependency of the IRS coverage, including the antenna directivity of the transmitter and propagation characteristics.

Wireless sensor and actuator networks (WSANs) are required to achieve both energy-efficiency and low-latency in order to prolong the network lifetime while being able to quickly respond to actuation commands transmitted based on the real-time sensing data. These two requirements are in general in a relationship of trade-off when each node operates with well-known duty-cycling modes: nodes need to make their radio interfaces (IFs) frequently active in order to promptly detect the communication requests from the other nodes. One approach to break this inherent trade-off, which has been actively studied in recent literature of wireless sensor networks (WSNs), is the introduction of wake-up receiver that is installed into each node and used only for detecting the communication requests. The main radio IF in each node is woken up only when needed, i.e., in an on-demand manner, through a wake-up message received by the wake-up receiver. In this paper, we introduce radio-on-demand sensor and actuator networks (ROD-SAN) where the concept of wake-up receiver is applied to realize on-demand WSANs. We first evaluate data collection rate, packet delivery latency, and energy-efficiency of ROD-SAN and duty-cycling modes defined in IEEE 802.15.4e by computer simulations. Then, we present our test-bed implementation of ROD-SAN including all protocols from the lowest layer of wake-up signaling to the application layer offering the functionalities of information monitoring and networked control. Finally, we show experimental results obtained through our field trial in which 20 nodes are deployed in an outdoor area with the scale of 450m × 200m. The numerical results obtained by computer simulations and experiments confirm the effectiveness of ROD-SAN to realize energy-efficient and high-response WSANs.