In this article, we show the development and implementation of the Internet of Maritime Things (IoMT) platform that supports long-range and high-rate communication for remote and online marine water quality monitoring. We first develop the IoMT device and gateway that interwork with international standard Internet-of-Things (IoT) software platforms for universal utilization in the field of marine monitoring. The proposed IoMT device and gateway play a role similar to the IoT device and gateway on land, but it is optimized for marine use with consideration of the practical characteristics of using sensors in the ocean. Furthermore, to realize long-range and high-rate communication between marine sensors and a control center (CC) on land, an automatic beam adjustment system for directional antennas is developed. This enables the use of directional antennas, which can support long-range and high-rate communication by automatically adjusting the beam angle between the antennas while exchanging their control information, despite the antennas being shifted by undulation from the sea. Finally, we conduct real sea experiments to identify the performances of the proposed IoMT platform in various environments. The results show that the proposed IoMT platform can collect data from marine sensors as well as high definition camera video from distances of 25 and 40 km to the CC on land under one-hop and two-hop transmissions, respectively. These results indicate that the proposed IoMT platform provides at least twofold performance improvement in terms of communication coverage under a similar data rate condition, compared with existing commercial wireless networks (e.g., LTE in the ocean) not utilizing satellite communications.

In smart grid, an advanced metering infrastructure (AMI) facilitates two-way communications between smart meters (SMs) and a meter data management unit (MDMU). We use cellular-assisted device-to-device (D2D) communications to connect a large number of SMs to the MDMU via cluster heads (CHs). Improper clustering can affect spectral efficiency. In addition, each SM and CH must be provided with a throughput guarantee for reliable smart grid operations. In this paper, we formulate a joint D2D cluster formation and radio resource allocation problem to maximize spectral efficiency in terms of uplink sum-rate, subject to constraints of guaranteeing a minimum throughput to each AMI entity. The optimization problem is a mixed-integer non-linear program, which is non-deterministic polynomial-time hard. For practical implementation and real-time decision making, we divide the original problem into four sub-problems: cluster formation, initial power allocation, channel allocation, and power allocation improvement, respectively. In the sub-problem for power allocation improvement, we exploit the properties of the difference of convex functions in finding the optimal transmission power. Simulation results confirm superiority of the proposed algorithm compared with other existing algorithms.

In this study, we first present a framework that jointly optimizes energy harvesting and information decoding for Internet of Things (IoT) devices, which are capable of simultaneous wireless information and power reception, in a smarty city. In particular, a generalized power-splitting receiver for IoT devices is designed, where each antenna in the receiver has an independent power splitter, unlike the existing works in which only one power splitter is employed regardless of the number of antennas in the receiver. Such a receiver design can provide a great degree of freedom to improve the network performance. Based on the presented framework, for each IoT device, we formulate an optimization problem whose objective is to maximize the harvested energy of each IoT device while satisfying its data rate requirement. To solve this problem, we propose a double-deep deterministic policy gradient-based online learning algorithm which enables each IoT device to jointly determine receive beamforming and power-splitting ratio vectors in real time. Furthermore, each IoT device can implement the proposed algorithm in a distributed manner using only its local channel state information. As such, cooperation and information exchange among the base stations and IoT devices are not necessary when performing the proposed algorithm at IoT devices. The extensive simulation results show the validity of the proposed algorithm.

This paper analyzes the coexistence performance of Wi-Fi and cellular networks conditioned on nonsaturated traffic in the unlicensed spectrum. Under the condition, the time-domain behavior of a cellular small-cell base station (SCBS) with a listen-before-talk (LBT) procedure is modeled as a Markov chain, and it is combined with a Markov chain which describes the time-domain behavior of a Wi-Fi access point. Using the proposed model, this study finds the optimal contention window size of cellular SCBSs in which total throughput of both networks is maximized while satisfying the required throughput of each network, under the given traffic densities of both networks. This will serve as a guideline for cellular operators with respect to performing LBT at cellular SCBSs according to the changes of traffic volumes of both networks over time.

In this study, the problem of dynamic channel access in distributed underwater acoustic sensor networks (UASNs) is considered. First, we formulate the dynamic channel access problem in UASNs as a multi-agent Markov decision process, wherein each underwater sensor is considered an agent whose objective is to maximize the total network throughput without coordinating with or exchanging messages among different underwater sensors. We then propose a distributed deep Q-learning-based algorithm that enables each underwater sensor to learn not only the behaviors (i.e., actions) of other sensors, but also the physical features (e.g., channel error probability) of its available acoustic channels, in order to maximize the network throughput. We conduct extensive numerical evaluations and verify that the performance of the proposed algorithm is similar to or even better than the performance of baseline algorithms, even when implemented in a distributed manner.

For the efficient use of radio resource, semi-persistent scheduling (SPS) has been adopted to reduce control overhead required for resource allocation. In this letter, for voice over internet protocol traffic, a lower bound on blocking probability and a closed-form maximum achievable control overhead reduction ratio are derived through a novel Markov chain model. Also, the performance evaluation of SPS under various resource assignment algorithms is provided on the blocking probability and overhead reduction ratio. Analytical expressions are validated through simulations.