By Ayfer Özgür, Ecole Polytechnique Fédérale de Lausanne, Faculté Informatique et Communications, Switzerland, firstname.lastname@example.org | Olivier Lévêque, University of California at Berkeley, Department of EECS, USA, email@example.com | David Tse, Ecole Polytechnique Fédérale de Lausanne, Faculté Informatique et Communications, Switzerland, firstname.lastname@example.org
Multi-hop is the current communication architecture of wireless mesh and ad hoc networks. Information is relayed from each source to its destination in successive transmissions between intermediate nodes. A major problem regarding this architecture is its poor performance at large system size: as the number of users in a wireless network increases, the communication rate for each user rapidly decreases. Can we design new communication architectures that significantly increase the capacity of large wireless networks?
In this monograph, we present a scaling law characterization of the information-theoretic capacity of wireless networks, which sheds some light on this question. We show that the answer depends on the parameter range in which a particular network lies, namely the operating regime of the network. There are operating regimes where the information-theoretic capacity of the network is drastically higher than the capacity of conventional multi-hop. New architectures can provide substantial capacity gains here. We determine what these regimes are and investigate the new architectures that are able to approach the information-theoretic capacity of the network. In some regimes, there is no way to outperform multi-hop. In other words, the conventional multi-hop architecture indeed achieves the information-theoretic capacity of the network. We discuss the fundamental factors limiting the capacity of the network in these regimes and provide an understanding of why conventional multi-hop indeed turns out to be the right architecture.
The monograph is structured as follows: In Section 2, we discuss the role of interference in wireless networks. We show that while current communication architectures are fundamentally limited by interference, new architectures based on distributed MIMO communication can overcome this interference limitation, yielding drastic performance improvements. Section 3 discusses the impact of power. We show that in power-limited regimes, distributed MIMO-based techniques are important not only because they remove interference but also because they provide received power gain. We identify the power-limited operating regimes of wireless networks and define the engineering quantities that determine the operating regime of a given wireless network. We show that unless the wireless network operates in a severely power-limited regime, distributed MIMO communication provides significant capacity gain over current techniques. Finally, in Section 4, we study how the area of the network, i.e., space, impacts the capacity of the network. This study enriches the earlier picture by adding new operating regimes where wireless networks can be moderately or severely space-limited. We see that unless the network is severely limited in space, distributed-MIMO-based communication continues to provide drastic improvements over conventional multi-hop.
The current communication architectures of wireless mesh and ad hoc networks are fundamentally limited by interference between simultaneous transmissions. This interference limitation leads to poor performance in large networks, where there are typically many source-destination pairs that want to communicate simultaneously: as the number of users in the network increases, the communication rate for each pair rapidly decreases. Given the increasing need to connect a massive number of wireless devices and to support various resource-intensive applications today, this poses a big challenge in the proliferation of such networks. Can we design new architectures that significantly increase the capacity of large wireless networks? Operating Regimes of Large Wireless Networks answers this question based on a scaling law characterization of the information-theoretic capacity of wireless networks. It shows that the information-theoretic capacity of wireless networks is much larger than the capacity of current architectures. In particular, the interference barrier limiting current performances can be surpassed with a combination of physical layer and architectural ideas and such new architectures can provide dramatic performance gains in large networks. Operating Regimes of Large Wireless Networks identifies the fundamental operating regimes and system parameters in wireless networks, clarifies the impacts of main limiting factors, such as interference, power and space, and suggests architectural guidelines for the design of optimal architectures. It aims to provide a basis for engineers in the field of wireless communications and networking to explore new directions and methods for designing the future of wireless networks.