Ultra-high bandwidth, negligible latency and seamless communication are envisioned as milestones that will revolutionize the way by which societies create, distribute and consume information. The remarkable expansion of wireless data traffic has advocated the investigation of suitable regimes in the radio spectrum to satisfy users’ escalating requirements and allow the exploitation of massive capacity and massive connectivity. To this end, the Terahertz (THz) frequency band (0.1-10 THz) has received noticeable attention in the research community as an ideal choice for scenarios involving high-speed transmission. As such, in this work, we present an up-to-date review paper to analyze key concepts associated with the THz system architecture. THz generation methods are first addressed by highlighting the recent progress in the devices technology. Moreover, the recently proposed channel models available for propagation at THz band frequencies are introduced. A comprehensive comparison is then presented between the THz wireless communication and its other contenders. In addition, several applications of THz communication are discussed taking into account various scales. Further, we highlight the milestones achieved regarding THz standardization activities. Finally, a future outlook is provided by presenting and envisaging several potential use cases and attempts to guide the deployment of the THz frequency band.

Evolving Internet-of-Things (IoT) applications often require the use of sensor-based indoor tracking and positioning, for which the performance is significantly improved by classifying the type of the surrounding indoor environment. This classification is of high importance since it leads to efficient power consumption, when operating the deployed IoT sensors. This paper presents a machine learning approach for indoor environment classification based on real-time measurements of the RF signal in a realistic environment. Several machine learning classification methods are explored including decision trees, support vector machine, and k-Nearest Neighbor using different RF features. Results obtained show that a machine learning approach based on weighted k-Nearest Neighbor method, that utilizes a combination of channel transfer function and frequency coherence function, outperforms the other methods in classifying the type of indoor environment with an accuracy of 99.3%. The predication time was found to be below 10's, which verifies that the adopted algorithm are successful candidates for real-time deployment scenarios.

This paper presents a high-sensitive planar microwave sensor based on the metamaterial technology, integrated with a microfluidic channel to be used for liquid sensing in a variety of biomedical applications. The sensor design comprises three cells of circular complementary split ring resonators (CSRRs) engraved on the ground plane of a dielectric substrate in a cascaded configuration. The sensing cells are excited with a time-varying electric field coupled from a planar microstrip-line (MTL) etched on the bottom side of the substrate. The proposed design of multiple coupled resonators would enhance the electric field intensity over a larger sensing region through the inter-resonator coupling, thus improving the overall sensitivity for detecting liquid samples of high permittivity and dielectric losses. The sensor exhibits a reject (band stop) filter behaviour with multiple resonances in the centimeter-wave band 1 – 6 GHz when loaded symmetrically with liquid samples. The sensor is fabricated on an FR4 dielectric substrate with a biocompatible microfluidic channel aligned appropriately over the sensing area to enable consistent and intact sensing of the liquid samples. The CSSR-sensor is numerically modeled and compared in performance to the single-cell CSRR for detecting small variations in the dielectric properties. The sensitivity, reliability, and repeatability of the proposed sensor are practically demonstrated by the in-lab measurements for different liquid samples using a Vector Network Analyzer (VNA) setup where distinctive resonant features (amplitude and frequency) are extracted from multiple resonance modes in the reflection and transmission responses. A high sensitivity is also demonstrated for monitoring glucose level variations in synthetic blood samples of concentrations (70 – 150 mg/dL) which are clinically-relevant to diabetes conditions. Beside its impressive capability in detecting small dielectric variations, the developed sensor features other favorable attributes including compact size, simple fabrication, affordable cost, non-ionizing nature, and minimal health risk or impact. Such key advantages could potentially promote the proposed sensor for integration with other microwave components in an embedded device for non-invasive monitoring of blood glucose for diabetes.

Several location-based services require accurate location information in indoor environments. Recently, it has been shown that deep neural network (DNN) based received signal strength indicator (RSSI) fingerprints achieve high localization performance with low online complexity. However, such methods require a very large amount of training data, in order to properly design and optimize the DNN model, which makes the data collection very costly. In this paper, we propose generative adversarial networks for RSSI data augmentation which generate fake RSSI data based on a small set of real collected labeled data. The developed model utilizes semi-supervised learning in order to predict the pseudo-labels of the generated RSSIs. A proper selection of the generated data is proposed in order to cover the entire considered indoor environment, and to reduce the data generation error by only selecting the most realistic fake RSSIs. Extensive numerical experiments show that the proposed data augmentation and selection scheme leads to a localization accuracy improvement of 21.69% for simulated data and 15.36% for experimental data.

Evolling Internet-of-Things (IoT) applications often require the use of sensor-based indoor tracking and positioning, for which the performance is significantly improved by identifying the type of the surrounding indoor environment. This identification is of high importance since it leads to higher localization accuracy. This letter presents a novel method based on a cascaded two-stage machine learning approach for highly-accurate and robust localization in indoor environments using adaptive selection and combination of RF features. In the proposed method, machine learning is first used to identify the type of the surrounding indoor environment. Then, in the second stage, machine learning is employed to identify the most appropriate selection and combination of RF features that yield the highest localization accuracy. Analysis is based on $k$ -Nearest Neighbor ( $k$ -NN) machine learning algorithm applied on a real dataset generated from practical measurements of the RF signal in realistic indoor environments. Received Signal Strength, Channel Transfer Function, and Frequency Coherence Function are the primary RF features being explored and combined. Numerical investigations demonstrate that prediction based on the concatenation of primary RF features enhanced significantly as the localization accuracy improved by at least 50% to more than 70%.

In this letter, we propose dual-mode chirp spread spectrum (DM-CSS) modulation for low-power wide-area networks. DM-CSS is capable of achieving a higher spectral efficiency (SE) relative to its counterparts, such as Long Range (LoRa) modulation. Considering the same symbol period, the SE in DM-CSS are augmented by: (i) simultaneously multiplexing even and odd chirp signals; (ii) using phase shifts of 0 and π radians for both even and odd chirp signals; and (iii) using either up-chirp or down-chirp signal. The SE increases by up to 116.66% for the same bandwidth and spreading factor relative to LoRa. We present a complete transceiver architecture along with non-coherent detection process. Simulation results reveal that DM-CSS is not only more spectral efficient but also more energy efficient than most classical counterparts. It is also demonstrated that DM-CSS is robust to phase and frequency offsets.