Skip to main content

Wireless with Increased Network Density, Antennas, Spectrum and Heterogeneity

Periodic Reporting for period 2 - WINDASH (Wireless with Increased Network Density, Antennas, Spectrum and Heterogeneity)

Reporting period: 2018-09-12 to 2019-09-11

"This action studies models for the design of better wireless communications in the future. The H2020 ""Digital Agenda for Europe” Flagship Initiative aims at providing all EU citizens with Internet speeds above 30Mbps and above 50% households above 100Mbps. To reach all citizens and achieve the target rates, it is imperative that wireless cellular services increase their capacity.

Wireless with Increased Network Density, Antennas, Spectrum and Heterogeneity (W.I.N.D.A.S.H.) comprises all means of capacity increase in the cellular industry.

Network Density refers to the number of base stations per square kilometer. All the users in the area covered by a tower, called a cell, share its resources. Operators increase tower density reducing cell area to improve per-user rates. However, this increases interference and even the models traditionally used to analyze networks must be modified.

Increasing the number of Antenna elements or effective area in an array generally enables radios to focus more signal power in the desired direction reducing interference in other directions or, alternatively, transmit different information in each antenna to increase rate. Citizens are familiar with 3 or 4 antenna WiFi-N routers, but in recent years the deployment of radios with hundreds of antennas is being considered.

Telecommunication operators acquire transmission licenses for portions of the electromagnetic Spectrum. The more “bandwidth”, the more capacity in the channel. Nowadays capacity is principally increased via the acquisition of new spectrum. Operators expect to start using the mmWave band (30-300GHz), bid TV broadcasters out of their 700-800MHz license, and increase the bandwidth in the 800-2100MHz region to 100MHz. Out of these, only mmWave offers over 1GHz of bandwidth and can support multiple Gbps rates.

Heterogeneity refers to the need to support devices of different characteristics. Small cells are deployed on top of, and without deactivating, the former large-area base stations. MmWave communications require massive antenna arrays to increase range. Increases in Density, Antennas and Spectrum have complex effects on each other, and on channel estimation and interference.

In the past wireless communications, increasing these four D.A.S.H. factors were less appealing due to technical difficultis. However, current cellular wireless systems are very saturated. This project explores wireless network performance with wider bandwidths, massive antenna array architectures, and increasingly dense heterogeneous topologies. Through a better understanding of these growth vectors, we will pave the way for the design of new wireless networks and new communications services for EU citizens."
We developed an energy efficiency analysis for efficient massive antenna array receivers comparing analog combining, low resolution digital combining and hybrid combining designs for mmWave radios. We found that hybrid schemes only have a marginally better energy efficiency than digital schemes at the cost of a significant drop in spectral efficiency, and when both factors are desired the digital option may be preferable.

We studied the inter-play between large bandwidths and large receiver antenna arrays in terms of channel estimation difficulties. We found that if the bandwidth scales faster than the square root of the number of receive antennas, the full increase in bandwidth cannot be exploited in Rayleigh fading channels where the receiver needs to estimate the channel state. As a consequence, as future wireless operators acquire more licensed spectrum, they will have to invest in the installation of a very large number of receiver antennas.

We developed performance models for an advanced channel estimation technique called “compressed sensing” (CS), which exploits the “sparsity” of mmWave channels due to blockage and absortion. In our analysis we characterized the performance of CS algorithms as estimators of mmWave channels, showing that certain properties of mmWave channels make them specially well suited for CS.

We have presented a theoretical analysis for the capacity scaling of large cellular networks where the number of base stations, antennas, bandwidth, and user density grow. Our results show that as the bandwidth keeps increasing there is a “saturation point” where the increase in bandwidth no longer results in an increased capacity. Using single hop protocols, the base station directly transmits to each user separately, and due to the long distance of this transmission this strategy saturates sooner, taking less advantage of the increased bandwidth. On the other hand, using multi-hop protocols, the base station only communicates with the closest users, who retransmit the information forward to other users close to them until the cell edge users are eventually reached over a path of multiple short transmissions. This strategy reduces per-transmission distance and enables the exploitation of more bandwidth.

Following this result, we developed scheduling models for multi-hop mmWave networks. This model enables the application of our information-theoretical result and to simulate the operation of real mmWave heterogeneous cellular networks with multi-hop relays. We also extended our scheduling model to mmWave networks with non-negligible interference using the Simulated Annealing optimization framework. Simulation results have shown that practical mmWave picocell multi-hop networks can be implemented with rates above 1Gbps per user.
Our receiver comparison improved the accuracy of energy consumption analyses in the literature, where only the power consumed by analog-to-digital converters had been taken into account.

Our wideband massive receive array capacity analysis unified two existing, but formerly separate, branches of theoretical work: the analysis of wideband channels with a fixed number of antennas, and the analysis of massive antenna systems with a fixed bandwidth.

The main novelty in our network scaling analysis was the introduction of the influence of bandwidth to the “capacity scaling laws” theoretical literature. Although it is well known that in a point-to-point link the capacity saturates as the bandwidth keeps increasing, ours is the first theoretical model for the characterization of the same effect on a cellular network.

In our scheduling model, we extended traditional scheduling models for 4G mobile systems that considered only single hop and cannot support the multi-hop network architecture we have found to be better for future networks. Moreover, we introduced extensions to existing scheduling models for mmWave multi-hop networks that take into account interference and support more versatile power allocation schemes at the transmitter.

Overall, our project finds that in the future it will be possible to increase Wireless Network capacity via Increased Density, Antennas, Spectrum and Heterogeneity, but to do so certain traditional wireless networking paradigms must change: numerous relay devices must be densely deployed for multi-hop, and regulatory spectrum licensing policies must take into account that in certain networks the acquisition of new spectrum licenses will no longer result in capacity increases as it has been taken for granted during the prior 30 years.

The wider societal implications of our work include more reliable and faster Internet services to citizens, enabling new or improved digital services.