Statistical characterization of an urban dual‐polarized MIMO LMS channel

Abstract

This paper presents a thorough statistical characterization of a dual-polarized multiple-input-multiple-output channel, based on a measurement campaign of 2 land mobile satellite link scenarios in urban environments. The received signal is decomposed into its large-scale and small-scale fading parts, which are separately evaluated and characterized. The large-scale fading can be well approximated by the Lognormal distribution, whereas the small-scale fading in line-of-sight condition can be characterized as a Rician channel with a strong direct component and high Kfactors. On the other hand, in nonline-of-sight cases, the fading can be approximated by Nakagami or Rician distribution with low K-factors. The cross-polar discrimination is found between 14.2 and 22.5 dB, whereas the cross-polar isolation is found between 12.8 and 21 dB, respectively. Finally, assessing the diversity performance, applying the maximal ratio combining technique, it is found that the left-hand circular polarized transmission outperforms the right-hand circular polarized transmission, providing gains up to 2.5 dB, especially in nonline-of-sight cases. The low-diversity gains indicate that beamforming would be preferable to be applied in the specific channel.KEYWORDS channel characterization, channel modeling, dual-polarized, multiple-input-multiple-output (MIMO) satellite measurements | INTRODUCTIONIn the years to come, new narrowband satellite communication technologies will allow for direct connection of millions of Internet of Things (IoT) devices in a resource efficient way. The ultimate success of global IoT coverage will depend on the active support of satellite networks providing L-band and S-band services. Satellite technology serves as a key enabler to transform IoT connectivity across industries and geographical borders and has the potential to play a key role in cases where sensors and actuators are distributed over a very wide area as well as in remote areas not served by terrestrial access networks. Indeed, the use of the satellite is found to be of great importance in some applications of IoT, like smart grid, environmental monitoring, and emergency management, because satellite communications have unique merits such as large-scale coverage, superb ability to support emerging communications services, and cost effectiveness for broadcast/multicast connectivity. 1-3 Bringing wide-area connectivity for the IoT by using satellite technology is therefore an attractive solution to complement terrestrial networks, allowing densification and coverage extension in remote areas. 4The proliferation of multiantenna technology and its adoption in terrestrial communication systems provided unprecedented capacity improvement and enhanced spectral efficiency. The applicability of multiple-input-multiple-output (MIMO) techniques is also an appealing solution in satellite communication systems with a very promising outlook. 5,6 Specifically, land mobile satellite (LMS) systems can take advantage of MIMO techniques and achieve substantially in...

Neural Network Aided Computation of Mutual Information for Adaptation of Spatial Modulation

“…The true MI with QPSK, 8PSK and 16QAM constellations of each realization of (γ, H) is calculated with a Monte Carlo simulation using (9) with 5, 000 realizations of the complex Gaussian noise w. We limit ourselves to these low order constellations, which are more likely to be used with SM. However, other constellations, like 64QAM, could be easily added to the system at the expense of increasing the time required for obtaining the dataset with Monte Carlo simulations -note the two summations over all the constellation symbols in (9).…”

“…The evaluation of the Mutual Information (MI) (9) can be interpreted as a non-linear mapping from the channel matrix H and the SNR γ to the MI. Multilayer Feedforward Neural Networks (MFNNs), well-known for their fitting capabilities of non-linear functions [29]- [30], will be used to estimate I T in (9). In particular, the MFNN to be employed, a one hidden layer network, is detailed in Fig.…”

“…The output of the network in Fig. 2 is the MI estimation for K different constellations, as an approximation of the true MI function (9). This depends on the channel matrix H and the SNR γ; in the following we will see how to pre-process these values for a better network performance.…”

“…The MI will be also affected by all the involved distances, as inferred from (9). For convenience, we put together the squared distances among all the pairs of supersymbols under the matrix D of size 2M × 2M , with the respective entries given by…”

Building entry loss and capacity evaluation of a UAV‐to‐indoor dual‐polarised MIMO channel

“…Since D′ ≫ Δx + w the difference of the two angles is negligible (lower than 1°) and the approximation is valid. To calculate rigorously the entry loss, only the co-polarised components (RR and LL) are taken into account, so as to exclude any polarisation mismatch between the co-and cross-polarised signals, which may reach up to 22.5 dB [16]. However, there is still a power imbalance between the co-polarised components attributed only to the polarisation rotation caused by the interaction with the building.…”