Signal Processing Challenges in Satellite Communication Networks

Prof. Björn Ottersten, University of Luxembourg, Luxembourg

Satellite communications provides an unprecedented coverage at low cost. However, satellite networks as a means of content delivery are meeting increased competition from terrestrial communication systems. Future satellite systems must provide cost efficient  and scalable services as part of a 5G network, for example, multimedia delivery, mobile communication services, two-way data communications, and backhaul. The efficient and reliable delivery of these services poses several challenges. We discuss some technical trends that are changing the design of satellite communication networks fundamentally.  With lessons learned from terrestrial communication networks, we highlight techniques that can be used to address some challenges facing satellite systems, including diversity techniques, interference mitigation, multi-user precoding, resource management, onboard processing, and cognitive satellite communications.

Biography
Website
Slides

Physical Layer Challenges When Designing the Tactile Internet

Prof. Gerhard Fettweis, Technische Universität Dresden – Vodafone Chair Mobile Communications Systems, Germany

The Tactile Internet requires to design communications systems which enable an end-to-end latency target on the order of 1ms. Due to routing, control and decision making, as well as packet handling, the resulting physical layer latency requirement is on the order of 100µs.
Many different modulation techniques are being discussed today for meeting this challenge, as e.g. OFDM, filtered OFDM, FBMC, and GFDM. The talk will address the different modulation techniques and compare them in the context of the Tactile Internet latency requirement.
An additional requirement appears when the category of mission critical applications of the Tactile Internet are examined, namely resilience of the wireless connection. An overview of techniques and measurement results shows that tough reliability requirements can be met by diversity combining of unreliable links.

Website
Slides

Massive MIMO: implementation issues and impact on network optimization

Prof. Giuseppe Caire, Technische Universität Berlin, Germany

Massive MIMO is a physical layer technique that builds on the well-known idea of multiuser MIMO and exploits the regime where the number of jointly processed infrastructure antennas is much larger than the number of spatially multiplexed data streams. This regime, combined with the possibility of learning the downlink channel state information from uplink pilot signals via TDD reciprocity. Due to the large multiplexing gain and array gain, the conventional wisdom of classical cellular systems optimization (indeed, the very concept of “cells” and user-base station association based on Voronoi tassellation of the coverage area) must be revised. At the same time, due to the “channel hardening” effect arising in high dimensional MIMO, many of the classical tasks such as opportunistic user scheduling, rate allocation, and user-cell association, are dramatically simplified and several notoriously difficult problems (NP-hard) become convex and suitable to decentralized low-complexity solutions with massive MIMO. In this talk we shall present some progress towards the hardware-efficient implementation of massive MIMO, exploiting key features of the channel structure that appear in high-dimension. Furthermore, we shall discuss the impact of massive MIMO at the level of network architecture optimization. In particular, we shall demonstrate hybrid digital analog implementation and semi-decentralized network optimization schemes that achieve very high system capacity with very limited complexity, both at the individual base station hardware level, and at the level of the overall system optimization.

Website
Slides

Massive, Ultra-Reliable, and Low-Latency Wireless: How to Communicate with Short Packets

Prof. Petar Popovski, Aalborg University, Denmark

Most of the recent advances in the design of high-speed wireless systems are based on information-theoretic principles that demonstrate how to efficiently transmit long data packets. However, the upcoming wireless systems, including 5G, will need to support novel traffic types that use short packets. For example, short packets represent the most common form of traffic generated by sensors and other devices involved in Machine-to-Machine (M2M) communications. Furthermore, there are emerging applications in which small packets are expected to carry critical information that should be received with low latency and ultra-high reliability. The design of current wireless systems relies on the assumption that the metadata (control information) is of negligible size compared to the actual information payload. Hence, transmitting metadata using heuristic methods does not affect the overall system performance. However, when the packets are short, metadata may be of the same size as the payload, and the conventional methods to transmit it may be highly suboptimal. This talk will provide insights in the communication-theoretic principles for communication with short packets. These principles are then applied to design communication protocols in several exemplary scenarios. The insights brought by these examples suggest that new approaches are needed to design of wireless protocols supporting short packets.

Website
Slides

Degrees of Freedom on Cached MIMO Interference Networks

Prof. Vincent Lau, Hong Kong University of Science and Technology, Hong Kong

There are two fundamental issues, namely the interference and the backhaul, associated with modern wireless networks. In this talk, we explore a new way of utilizing cache in the wireless network called the PHY caching to mitigate interference with limited backhaul. We first give an overview of the PHY caching and explain the key design challenges. Next, we discuss the performance benefit of PHY caching in MIMO interference network in terms of the “Degrees of Freedom”. Specifically, we first establish a converse results on the maximum DoF of Cached MIMO interference channels. We then proposed a scheme to achieve the DoF results. We also discuss how the DoF depends on various system parameters such as the content popularity distribution and cache capacity.

Biography
Website
Slides

Vehicle-to-X communication using millimeter waves

Prof. Robert Heath, The University of Texas at Austin, USA

Vehicles are becoming more intelligent and automated. To achieve higher automation levels, vehicles are being equipped with more and more sensors. High data rate connectivity seems critical to allow vehicles exchanging all these sensor data to enlarge their sensing range and make better safety related decisions. Current solutions for vehicular communications though do not support  the gigabit-per-second data rates required to exchange raw sensor data between the vehicles or between the vehicles or the infrastructure. This presentation makes the case that millimeter wave communication is the only viable approach for high bandwidth connected vehicles. The motivation and challenges associated with using mmWave for vehicle-to-vehicle and vehicle-to-infrastructure applications are highlighted. Examples from recent work are provided including new notions of coherence time and innovative architectural concepts like radar-aided communication.

Website
Slides

Team approaches in decentralized device coordination

Prof. David Gesbert, EURECOM, France

Future wireless networks face serious challenges related to the ever growing popularity of mobile applications and the increasing diversity of connected objects. Two fundamental approaches, namely network-centric cloud-based solutions and device-centric solutions, offer conflicting design philosophies to respond to the challenge. From a decisional and optimisation persepective, device-centric designs rely on the device’s own local intelligence and data gathering to help improve overall network performance. The concept of device cooperation is instrumental to enabling a form of collective intelligence arising from the devices themselves.

Device cooperation can aim at a number of goals such as interference management, spectrum allocation, coordinated multipoint transmissions, feedback design, etc. When the communications between the devices is itself (rate or delay) limited, the devices can at best reach robust yet suboptimal decisions that operate on the basis of local noisy feedback. Taking the example of spatial multiplexing across distant transmitters, we emphasize concepts such as the price of distributedness, and the development of precoding methods that are robust to distributed channel state information.

Slides

5G: Can we make it by 2020?

Prof. Mérouane Debbah, Centrale Supélec – Huawei France, France

If we look at some of the goals of 5G versus where we are today,  we can see the gap that has to be bridged over the next few years. The goal of 1ms latency is nearly 50x better than current LTE systems. In order  to go from 100Mbps per user to 10Gbps,  we need 100x the throughput per connection. The current 10,000 connections per square kilometer needs to increase to 1Million connections, which corresponds to 100x increase in density. Reliable communications today with LTE top out about 350km/h and we expect to bring that up by 1.5x to 500km/h . Finally,  the current core networks and backhaul/fronthaul are inflexible with wasted pools of bandwidth. The introduction of SDN/NFV will allow much better ability to chop up and virtualize the network resources for lower operational costs and capital costs and much greater flexibility.

In this talk, we will give an update on standardization process and the challenges ahead for a true deployment by 2020.
Slides

Cognition-Based Networks: A New Perspective on Network Optimization Using Learning and Distributed Intelligence

Prof. Michele Zorzi, University of Padova, Italy

In response to the new challenges in the design and operation of communication networks, and taking inspiration from how living beings deal with complexity and scalability, in this talk we introduce an innovative system concept called COgnition-BAsed NETworkS (COBANETS). The proposed approach develops around the systematic application of advanced machine learning techniques and, in particular, unsupervised deep learning and probabilistic generative models for system-wide learning, modeling, optimization, and data representation. Moreover, in COBANETS we propose to combine the learning architecture with the emerging network virtualization paradigms, which make it possible to actuate automatic optimization and reconfiguration strategies at the system level, thus fully unleashing the potential of the learning approach. Compared to past and current research efforts in this area, the technical approach depicted in this paper is deeply interdisciplinary and more comprehensive, calling for the synergic combination of expertise of computer scientists, communications and networking engineers, and cognitive scientists, with the ultimate aim of breaking new ground through a profound rethinking of how the modern understanding of cognition can be used in the management and optimization of telecommunication networks.

Biography
Slides