Evolving Internet-of-things 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 radio frequency (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 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%.

Indoor Localization for IoT Using Adaptive Feature Selection: A Cascaded Machine Learning Approach

Radiation from printed millimeter-wave antennas integrated in mobile terminals is affected by surface currents on chassis, guided waves trapped in dielectric layers, superstrates, and the user’s hand, making mobile antenna design for 5G communication challenging. In this paper, four canonical types of printed 28-GHz antenna elements are integrated in a 5G mobile terminal mock-up. Different kinds of terminal housing effects are examined separately, and the terminal housing effects are also diagnosed through equivalent currents by using the inverse source technique. To account for the terminal housing effects on a beam-scanning antenna subarray, we propose the effective beam-scanning efficiency to evaluate its coverage performance. This paper presents the detailed analysis, results, and new concepts regarding the terminal housing effects, and thereby provides valuable insight into the practical 5G mobile antenna design and radiation performance characterization.

Radiation Performance Analysis of 28 GHz Antennas Integrated in 5G Mobile Terminal Housing

Both massive multiple-input multiple-output (MIMO) technology and use of frequencies in the range 24100 GHz are considered essential for upcoming 5G systems. Measurements of the new type of channels are needed, but this is challenging due to the large number of channels in massive MIMO and the short wavelength in the 24100 GHz channels (so-called mmWave channels). This article describes a sounder system capable of measurements in both types of channels. While the same sounder system supports simultaneous massive MIMO and mmWave channels, the number of channels in each band and in total is in practice limited. For massive MIMO in the 3006000 MHz band alone, arrays with up to 128 receiver (Rx) elements are possible, receiving from 16 independent mobile transmitter (Tx) antennas, where all channels are measured within 1.3 ms and in a 200 MHz bandwidth. To the authors' knowledge, this is the first sounding system with this number of channels and a speed necessary for measuring the dynamic channels in typical application scenarios. For mmWave channels, the sounder operates in the range 1840 GHz. For these bands up to 2 Tx elements and 16 Rx elements can be measured. In addition, the article describes some measurements in both a massive MIMO setup in an indoor sports arena and an mmWave setup with a

/pdf/a-channel-sounder-for-massive-mimo-and-mmwave-channels-5h50ql9pnx.pdf

A Channel Sounder for Massive MIMO and MmWave Channels

The recently proposed channel model by the Third Generation Partnership Project for fifth generation New Radio (5G-NR) systems at millimeter- wave carrier frequencies in TR38.901 provides a good starting point for understanding the impact of propagation, material penetration, and blockage in cellular access systems. With this backdrop, the focus of this article is on hand and body blockage which have a significant bearing on user equipment (UE) design and in the practical viability of millimeter-wave systems. Building on TR38.901, we first study hand and body blockage loss with experiments using a 28 GHz experimental prototype that emulates a 5G base station and a form-factor UE. These measurement estimates show that while blockage losses can be substantial, they are significantly lower than the agreed model in TR38.901. We then study the implications of these revised estimates on spherical coverage with the UE. For this, we consider two form-factor UE designs: a face design and an edge design. Our studies show that while both designs are generally comparable in terms of spherical coverage, practical design considerations favor the edge design over the face design. We finally report on the timescales at which blockage events occur and show that given the symbol periods in 5G-NR, it is possible to implement mitigation strategies to overcome blockage (e.g., switch beams at the UE from a blocked path to an unblocked path).

Spatio-Temporal Impact of Hand and Body Blockage for Millimeter-Wave User Equipment Design at 28 GHz

Directional scanning measurement is the dominant strategy taken to measure channel spatial profiles for millimeter wave (mmWave) frequency bands. The focus of the paper is on wideband channel spatial profile reconstruction and angular sampling interval (ASI) selection for directional scanning measurement. We propose to employ the trigonometric interpolation technique to reconstruct the spatial channel profile and to use Nyquist sampling theorem and spherical wave mode theory to determine the ASI. Our analysis demonstrated that the half-power beam width (HPBW) of the antenna is a good approximation for the ASI setting to ensure accurate directional scanning measurements. The reconstructed power-angle-delay profile (PADP), composite power angular spectrum (PAS), delay and angle characteristics match well with those of the target in the measured data when the ASI is set to the antenna HPBW.

On Angular Sampling Intervals for Reconstructing Wideband Channel Spatial Profiles in Directional Scanning Measurements

For mobile systems involving hand-held devices, the influence of the user on system performance has to be considered. Extensive studies below 6 GHz have demonstrated large effects on system performance. However, the impact of user influence at potential frequency bands for upcoming 5G mobile networks is still to be investigated. This paper investigates how the user affects the performance of a 5G handset mock-up. The user impact is studied by channel sounding in an indoor scenario, with and without the presence of different users. The mock-up handset has a uniform linear array of receive (Rx) antennas operated at 21.5 GHz. A dual-polarized horn antenna with a wide beamwidth is transmit antenna and a fast channel sounder is used, allowing for dynamic and realistic channels. The results show that the mean influence of the user on the power varies considerably depending on the scenario, with more than 12 dB loss in some cases, while a gain of 4 dB is seen in other. An important finding is that the mean power among the seven Rx branches may be very different. Branch power ratios in the typical range of 2–10 dB were found, depending on the user and scenario.

/pdf/measured-21-5-ghz-indoor-channels-with-user-held-handset-4a48jaaum3.pdf

Measured 21.5 GHz Indoor Channels With User-Held Handset Antenna Array

In this letter, we propose a novel map-free indoor localization method using ultrawideband large-scale array systems in high-frequency bands The proposed localization method comprises two stages, namely a channel-estimation stage and a target-localization stage Due to the large array aperture, the locations of the scatterers associated to the multipath components can be estimated with the spherical wavefront model The estimated scatterers are further used as virtual anchors to estimate the location of the target through trilateration Since the scatterer locations are obtained from channel measurements, the map of the environment is not needed for localization The proposed method is also assessed with measurements conducted in a cluttered indoor environment with line-of-sight (LoS) and non-LoS scenarios Results show that the proposed algorithm attains good localization accuracy in both scenarios

/pdf/a-map-free-indoor-localization-method-using-ultrawideband-qt5cp6whd5.pdf

A Map-Free Indoor Localization Method Using Ultrawideband Large-Scale Array Systems

In this paper, we present validation of a method for constructing a virtual omnidirectional antenna in the azimuth plane. The virtual omnidirectional antenna utilizes a combination of data from directive horn antennas. The aim is to utilize the high gain of the horn antenna to improve the dynamic range of channel sounding measurements conducted in the centimeter and millimeter wave bands. The resulting complex impulse response from the virtual omnidirectional antenna is used to find the power-delay-profile (PDP). This is then compared to measurements conducted at the same time using a real omnidirectional antenna. The validation shows that the synthesized omnidirectional is capable of predicting main components and the slope of the PDF. Further, it is shown that by choosing angular sampling steps corresponding to the half power beam width (HPBW) of the used antenna similar power levels can be achieved.

/pdf/validation-of-emulated-omnidirectional-antenna-output-using-2x6srcv5cq.pdf

Validation of emulated omnidirectional antenna output using directive antenna data

With the future LTE auction for TV white spaces at 600 MHz, there is a strong need for efficient handset antennas operating at very low frequencies. This paper proposes a tunable antenna design for the low bands of LTE. In this design, not only the antenna is tuned but also the resonance of the board, thanks to using a tunable parasitic. The resulting dual-resonant antenna exhibits a peak total efficiency of −4 dB at 600 MHz.

/pdf/combining-antenna-and-ground-plane-tuning-to-efficiently-3r7z3uu38e.pdf

Johannes Hejselbak

Papers

Measured 21.5 GHz Indoor Channels With User-Held Handset Antenna Array

A Map-Free Indoor Localization Method Using Ultrawideband Large-Scale Array Systems

Validation of emulated omnidirectional antenna output using directive antenna data

Combining antenna and ground plane tuning to efficiently cover Tv white spaces on handsets