Outdoor-to-indoor signal propagation poses significant challenges to millimeter-wave link budgets. To gain insight into outdoor-to-indoor millimeter-wave at 28GHz, we conducted an extensive measurement campaign consisting of over 2,200 link measurements in West Harlem, New York City, covering seven highly diverse buildings. A path loss model constructed over all measured links shows an average of 30dB excess loss over free space at distances beyond 50m. We find the type of glass to be the dominant factor in outdoor-to-indoor loss, with 20dB observed difference between grouped scenarios with low-and high-loss glass. Other factors such as the presence of scaffolding, tree foliage, or elevated subway tracks, as well as difference in floor height are also found to have a 5–10dB impact. We show that for urban buildings with high-loss glass, outdoor-to-indoor downlink capacity up to 400Mb/s is supported for 90% of indoor customer premises equipment by a base station up to 40m away. For buildings with low-loss glass, such as our case study covering multiple classrooms of a public school, downlink capacity over 2.8/1.4Gb/s is possible from a base station 57/133m away within line-of-sight. We expect these results to help inform the planning of millimeter-wave networks targeting outdoor-to-indoor deployments in dense urban environments, as well as provide insight into the development of scheduling and beam management algorithms.

Millimeter-Wave (mmWave) has great potential to provide high data dates given its large available bandwidth, but its severe path loss and high propagation sensitivity to different environmental conditions make deployment planning particularly challenging. Traditional slope-intercept models fall short in capturing large site-specific variations due to urban clutter, terrain tilt or foliage, and ray-tracing faces challenges in characterizing mmWave propagation accurately with reasonable complexity. In this work, we apply machine learning (ML) techniques to predict mmWave path loss on a link-to-link basis over an extensive set of 28 GHz field measurements collected in a major city of USA, with over 120,000 links from both urban and suburban scenarios, with over 40 dB variation for links at similar distances. Either raw environmental profile (terrain+clutter) of each link or 8 selected expert features are used to either directly predict path loss via regression-based approaches or predict the best performing option out of a pool of theoretical/empirical propagation models. Our evaluation shows that Lasso regression provides the best path loss prediction with a performance (RMSE 8.1 dB) comparable to the per-site slope-intercept fit (RMSE 8.0 dB), whereas model selection method achieves 8.6 dB RMSE, both are significantly better than the best a posteriori 3GPP model (UMa-NLOS, 10.0 dB).

In the US, people spend 87% of their time indoors and have an average of four connected devices per person (in 2020). As such, providing indoor coverage has always been a challenge but becomes even more difficult as carrier frequencies increase to mmWave and beyond. This paper investigates the outdoor and outdoor-indoor coverage of an urban network comparing globally standardized building penetration models and implementing models to corresponding scenarios. The glass used in windows of buildings in the grid plays a pivotal role in determining the outdoor-to-indoor propagation loss. For 28 GHz with 1 W/polarization transmit power in the urban street grid, the downlink data rates for 90% of outdoor users are estimated at over 250 Mbps. In contrast, 15% of indoor users are estimated to be in outage, with SNR <−3 dB when base stations are 400 m apart with one-fifth of the buildings imposing high penetration loss (∼ 35 dB). At 3.5 GHz, base stations may achieve over 250 Mbps for 90% indoor users if 400 MHz bandwidth with 100 W/polarization transmit power is available. The methods and models presented can be used to facilitate decisions regarding the density and transmit power required to provide high data rates to majority users in urban centers.

Outdoor-to-indoor (OtI) signal propagation further challenges link budgets at millimeter-wave (mmWave). To gain insight into OtI mmWave at 28 GHz, we conducted an extensive measurement campaign consisting of over 2,000 link measurements in West Harlem, New York City, covering seven highly diverse buildings. A path loss model constructed over all links shows an average of 30 dB excess loss over free space at distances beyond 50 m. We find the type of glass to be the dominant factor in OtI loss, with 20 dB observed difference between clustered scenarios with low- and high-loss glass. Other factors, such as difference in floor height, are found to have an impact between 5--10 dB. We show that for urban buildings with high-loss glass, OtI data rates up to 400 Mb/s are supported for 90% of indoor users by a base station (BS) up to 49 m away. For buildings with low-loss glass, such as our case study covering multiple classrooms of a public school, data rates over 2.8/1.4 Gb/s are possible from a BS 68/175 m away when a line-of-sight path is available. We expect these results to be useful for the deployment of OtI mmWave networks in dense urban environments and the development of scheduling and beam management algorithms.

Accurate path gain models are critical for coverage prediction and radio frequency (RF) planning in wireless communications. In many settings irregular terrain induces blockages and scattering making it difficult to predict the path gain. Current solutions are either computationally expensive or slope-intercept fits that do not capture local deviations due to terrain variation, leading to large prediction errors. We propose to use machine learning to learn path gain based on terrain elevation as features. We implement different neural network architectures with dense and convolutional layers that could include effects difficult to describe with traditional models (e.g. back scatter). We test our framework on an extensive set of measured path gain data and consistently predict with 5 dB Root Mean Squared Error, an 8 dB improvement over traditional slope-intercept solutions.

Delivering Gbps high user rate over long distances (~1 km) is challenging, and the abundant spectrum available in millimeter wave band cannot solve the challenge by its own due to the severe path loss and other limitations. Since it is economically challenging to deploy wired backhaul every few hundred meters, relays (e.g., wireless access points) have been proposed to extend the coverage of a base station, which has wired connection to the core network. These relays, deployed every few hundred meters, serve the users in their vicinity and are backhauled to the base station through wireless connections. In this paper, the wireless-relayed backhaul design has been formulated as a topology-bandwidth-power joint optimization problem, and the influence of path loss, angular spread, array size, and RF power limitation on the user rate has been evaluated. It has been shown that for a linear network deployed along the street at 28 GHz, when high joint directional gain (50 dBi) is available, 1 Gb/s user rate within cell range of 1 km can be delivered using 1.5 GHz of bandwidth (using single polarization antennas). The user rates drop precipitously when joint directional gain is reduced, or when the path loss is much more severe. When the number of RF chains is limited, the benefit of larger arrays will eventually be surpassed by the increased channel estimation penalty as the effective beamforming gain saturates owing to the channel angular spread.

Theory of normal mode propagation in a line-of-sight street scenario is extended to include propagation into buildings through coupling to a diffuse indoor field. Signal strength predictions are in close agreement with measurements, producing 2 dB and ≤ 3.5 dB root-mean-square model-data difference, in line-of-sight and outdoor-indoor scenarios, respectively. The full 3-D field model predicts actual signal directions and antenna correlations as a function of range, important for evaluating performance of directional antennas and spatial diversity. Only minimal description of the environment is needed, i.e., street width and representative building wall properties, without any interior details.

Propagation theory is developed to get analytic expressions for both guided and diffusely scattered 3-D fields in common indoor environments, with access points (APs) placed along a hallway and mobile nodes in the rooms. Resulting predictions agree with measurements of both path loss and single-user multiple-input-multiple-output link capacity. In particular, the model reproduces the observed reduction in normalized channel capacity with increasing range. This effect, occurring in a common indoor environment, can be beneficial. Interfering APs, which tend to be further away, occupy channel dimensions corresponding to low-order hallway modes, causing total interference to concentrate in a small subset of the channel dimensions, in a manner reminiscent of interference alignment. The channel for the desired signal has a greater number of effective degrees of freedom (EDOFs) due to a shorter distance from its serving AP. The disparity in the number of EDOFs between the signal and the interference leads to higher rates, which can be further increased through water filling applied individually to each link, with no interlink coordination. The resulting rates are on the order of 30% greater than would be anticipated in range-independent scattering models.

Heterogeneous cellular networks are composed of a mixture of macro and small (pico) cells embedded in the macro network. A greedy strategy is proposed to maximize network capacity. Picos placement and either macro or pico transmit power are optimized jointly, as picos are deployed. The scaling of area spectral efficiencies (ASE) as pico densities are increased is studied. Median ASEs grow linearly with density and are comparable to cell-splitting given fixed pico power -20dB EIRP relative to the macro. The effectiveness of the pico at such low power is due in part to the improvement in area coverage of an omni vs. a 3 sectored antenna.

Diffusion theory is found to provide a simple yet accurate expression for the average received radio power loss indoors in non-line-of-sight propagation. Effective specific absorption is the sole parameter. Agreement with extensive indoor power measurements is found at all ranges. Use of the standard power-law to describe the data is found to lead to power exponents that vary from near 2 at short ranges to around 11 at 100 m. Extending the diffusion model to account for variation in scatterer density provides a stochastic mechanism explaining the log-normal nature of observed average power variations, usually termed "shadow fading". The standard deviation of power variation is found to increase with range (as r1/2at short ranges), in agreement with measurements and unlike accepted models. Exponential absorption in diffusion is determined to have a large impact on the signal-to-interference-and noise behavior, materially affecting conclusions on the value of increasing cell density as a method of increasing wireless network capacity.

We describe a method for estimating achievable data rates in urban microcells using multiple-input/multiple-output (MIMO) techniques. Specifically, we use site maps and a versatile ray-tracing tool to compute MIMO gain matrices as a function of terminal location; and we use these matrices to determine achievable rates for various MIMO transmission modes (spatial multiplexing, beamforming, and diversity). Numerical results are generated for specific paths in Boston and Manhattan, though our results are shown to be fairly insensitive to neighborhood or city. We also show that, in urban microcells, data rate prediction using site-specific ray tracing is more informative than using stochastic models; and that adaptive switching among MIMO transmission modes as a terminal moves along its trajectory can help sustain high data rates. A new mode-switching algorithm is proposed that requires switching rates lower than those for the optimal scheme by a factor greater than 10, with little loss in average data rate. We also propose a novel software algorithm for optimally placing microcell bases. For a Manhattan neighborhood of area 0.5 km2, we find that full coverage can be obtained using only 5 bases, and that the highest total throughput is achieved using a frequency reuse factor of 1.

It has been demonstrated that base station cooperation can reduce co-channel interference (CCI) and increase cellular system capacity. In this work we consider another approach by dividing the system into microcells through denser base station deployment. We adopt the criterion to maximize the minimum spectral efficiency of served users with a certain user outage constraint. In a two-dimensional hexagon array with homogeneous microcell structure, under the proposed propagation model denser base station deployment outperforms suboptimal cooperation schemes (zero-forcing) when the density increases beyond 3 - 12 base stations per km2, the exact value depending on the rules of outage user selection. However, close- to-optimal cooperation schemes (zero-forcing with dirty-paper- coding) are always superior to denser deployment. Performance of a hierarchial cellular structure mixed with both macrocells and microcells is also evaluated.

Wireless systems exploiting multi-user diversity require accurate channel state information. In this paper we examine channel prediction as a method to improve the performance of these systems. We model the outdoor channel by the Rician model, which combines diffuse spectra with a single specular component. Prediction accuracy is determined for the Rician model given various ratios (K factors) of specular to diffuse power. Simulation results show that accurate channel prediction further than one wavelength can be achieved at 30 dB SNR when the channel K factor is greater than ten. The predictor was evaluated on measured channels, and the Rician model was found to be appropriate as prediction intervals of the measured data are similar to those of synthetic data with same K factor. Using measurements some results regarding channel stationarity and optimal filter order are presented. We show for a proportionally fair scheduled downlink packet system that the predictor reduces fade margin at fixed packet outage leading to an improvement in throughput.

The wireless communication channel between vehicular nodes in an ad-hoc network was assessed in a comprehensive measurement and modeling effort. Nodes were equipped with roof mounted azimuthally omnidirectional antennas, and measurements were taken at both short (line-of-sight) and long ranges, at a center frequency of 2.5 GHz. The median MIMO capacity with 8 transmitters and 10 receivers (8 times10) was found to be about five times the corresponding 1times1 SISO capacity, and three times the corresponding 1times10 SIMO capacity. Wideband soundings of such channels were conducted, where the median rms delay spread was found to be 0.6 mus. Measurements, spatial correlation scales, an empirical model of pathloss, and cross- polarization properties of peer-to-peer channels in a rural area are reported.

Propagation of radio signals from a base above clutter, such as buildings and trees to a mobile immersed in clutter, is treated theoretically by accounting for random diffuse scattering at the mobile. Closed form expressions are derived for path gain as well as for angular spectrum at the base. The resulting predictions are in close agreement with widely accepted models and empirical results. The angular spectrum at the base is found to be Lorentzian of width close to that reported for urban measurements in Aarhus. Small-scale fading and distance-depedent loss are treated in a unified way, as opposed to the heuristic methodologies, which formulate them as separate factors.

Channel prediction is an appealing approach to mitigate channel mismatch in wireless downlink packet systems. Channel mismatch arises due to the delay in the feedback loop. Both synthetic and measured channels were used to assess the accuracy of channel prediction for various ratios of specular to diffuse power. The power spectrum was estimated rather than assumed known. It has been found that high ratios (K factors) greater than ten are necessary for prediction greater than one wavelength. K factors of measured channels gave prediction accuracy similar to that of the synthetic channel given by the same K factor.

It has been reported that the number of transmitters that can be used beneficially in a multiple-input multiple-output (MIMO) system is limited by the coherence time of the channel due to difficulties with channel estimation at the receiver. Furthermore, rapid channel fluctuations degrade the performance of feedback schemes that would otherwise increase the information throughput that may be achieved. These impediments either reduce achievable capacity or impose an effective "speed limit" on the mobile, above which the effective throughput is reduced. In this paper, the signal incident at the mobile is represented as a sum of plane waves. The channel transfer matrix is found to factor into time-dependent and time-independent parts. A processing method is proposed whereby the received signals are first preprocessed so as to produce signals that fluctuate on a much slower time scale. The preprocessing consists of beam forming, followed by Doppler compensation for signals received on each beam. Both operations are nonsingular and do not alter the capacity of the MIMO channel. Beam forming effectively partitions the angular spectrum at the mobile, with each partition suffering a smaller Doppler spread, resulting in a "slowed down" channel fluctuation due to mobile motion. The coherence time of the preprocessed channel is found to increase by a factor on the order of the number of mobile receive antennas. In the limit of an infinite number of mobile receive antennas, the MIMO channel is shown to become static. These results remove constraints imposed by training requirements on the number of transmit antennas and increase allowed vehicle speeds. The "slowed down" channel characteristics may also be fed back to the transmitter, allowing an increase in information throughput.

This paper presents a generalized multiple-input multiple-output (MIMO) channel modeling technique for link level and system level simulations. The model combines the correlation approach and wave superposition approach to achieve both accuracy and efficiency. The spatial, temporal, and frequency dispersions of the MIMO channels are implicitly modeled based on any given statistics. Polarizations, channel transitions, uplink and downlink channels, and keyhole-pinhole channels are modeled as optional modules. Theoretical justifications as well as experimental verification of the model are also presented. The proposed model can be applied to system simulations for MIMO as well as other adaptive antenna applications.

Narrowband multiple-input-multiple-output (MIMO) measurements using 16 transmitters and 16 receivers at 2.11 GHz were carried out in Manhattan. High capacities were found for full, as well as smaller array configurations, all within 80% of the fully scattering channel capacity. Correlation model parameters are derived from data. Spatial MIMO channel capacity statistics are found to be well represented by the separate transmitter and receiver correlation matrices, with a median relative error in capacity of 3%, in contrast with the 18% median relative error observed by assuming the antennas to be uncorrelated. A reduced parameter model, consisting of 4 parameters, has been developed to statistically represent the channel correlation matrices. These correlation matrices are, in turn, used to generate H matrices with capacities that are consistent within a few percent of those measured in New York. The spatial channel model reported allows simulations of H matrices for arbitrary antenna configurations. These channel matrices may be used to test receiver algorithms in system performance studies. These results may also be used for antenna array design, as the decay of mobile antenna correlation with antenna separation has been reported here. An important finding for the base transmitter array was that the antennas were largely uncorrelated even at antenna separations as small as two wavelengths.

In this paper, we discuss some of the most basic architectural superstructures for wireless links with multiple antennas: M at the transmit site and N at the receive site. Toward leveraging the gains of the last half century of coding theory, we emphasize those structures that can be composed using spatially one dimensional coders and decoders. These structures are investigated primarily under a probability of outage constraint. The random matrix channel is assumed to hold steady for such a large number of M-dimensional vector symbol transmission times, that an infinite time horizon Shannon analysis provides useful insights. The resulting extraordinary capacities are contrasted for architectures that differ in the way that they manage self-interference in the presence of additive receiver noise. A universally optimal architecture with a diagonal space-time layering is treated, as is an architecture with horizontal space-time layering and an architecture with a single outer code. Some capacity asymptotes for large numbers of antennas are also included. Some results for frequency selective channels are presented: It is only necessary to feedback M rates, one per transmit antenna, to attain capacity. Also, capacity of an (M,N) link is, in a certain sense, invariant with respect to signaling format.

This paper presents analysis for spatial and temporal variations of multiple-input-multiple-output (MIMO) channels at the mobile. The channel study is based on the narrowband measurements at 2.11 GHz in Manhattan, New York with 16 transmitting antennas and 16 receiving antennas. Doppler spread and angle of arrival (AOA) are derived from the temporal correlation of field components. The results show that the AOA at the mobile is not uniformly distributed. The restricted AOA distribution results in approximately twice the correlation distance and correlation time than the predicted values from the Jakes (1974) model. The measured median coherence time is at least a few seconds for stationary channels, and 90 ms at a mobile pedestrian speed of 3 km/hr. These observations hold for both vertically and horizontally polarized antennas. The measured median RMS angular spread at the mobile is 22.5/spl deg/ for horizontally polarized antennas and 25.5/spl deg/ for vertically polarized antennas.

Narrowband MIMO measurements using 16 transmitters and 16 receivers at 2.11 GHz were carried out in Manhattan. High capacities were found for full as well as smaller array configurations, all within 80% of the fully scattering channel capacity. It was found that two streets which might be considered near LOS, had different capacities, one about 45% of and the other about 75% of the fully scattering channel capacity.

The paper presents a multiple-input-multiple-output (MIMO) channel modeling technique for link level and system level simulations. The classical approach of wave superposition is adopted to generate MIMO channels that are statistically consistent in delay, frequency and angle domains. The paper proposes an efficient way to generate four dimensional MIMO channel transfer matrices based on standard channel statistics such as power delay profile (PDP) and power azimuth spectrum (PAS).

Multielement system capacities are usually thought of as limited only by correlations between elements. It is shown here that degenerate channel phenomena called "keyholes" may arise under realistic assumptions which have zero correlation between the entries of the channel matrix H and yet only a single degree of freedom. Canonical physical examples of keyholes are presented. For outdoor environments, it is shown that roof edge diffraction is perceived as a "keyhole" by a vertical base array that may be avoided by employing instead a horizontal base array.

We demonstrate the capacity growth of multiple-element antenna arrays (MEAs) in a realistic propagation environment using WiSE, an experimental ray tracing tool. WiSE is used to construct the channel response for MEAs operating at 1.9 GHz for two situations: (a) (16-16)-MEAs located inside an office building and (b) (4, 4)-MEAs located in an outdoor fixed wireless loop. We define effective degrees of freedom (EDOFs) as parallel spatial modes of transmission for an MEA. We quantify the increase in both the number of EDOFs and capacity with transmit power, received SNR, and antenna spacing. More EDOFs are present when receiving MEAs are physically closer to the transmitting MEA, regardless of the scattering effect.