Wireless network coding: the network aware PHY layer
Prof. Jan Sykora
Czech Technical University in Prague
Prof. Alister Burr
University of York, U.K.
It is becoming widely accepted that the most significant future developments in the physical layer of wireless communication systems will not take place in the PHY layer of individual communication links, but rather in the context of complete wireless networks. Over the past decade or so there have been significant developments in network information theory which have shown that very significant overall performance gains are available compared to the conventional paradigm in which PHY techniques are applied to individual links only, leaving network aspects to be dealt with only at higher layers of the protocol stack. One such new research field is network coding, in which coding techniques are applied to multiple data streams at intermediate nodes in a network, rather than only to individual streams on single links. This can exploit network topology to significantly improve throughput in multiuser networks. However in its original form it operates at the level of data streams, rather than signal waveforms, and hence is not well suited to the inherently broadcast nature of wireless networks. Physical layer or wireless network coding (WNC)allows it to be applied directly to wireless networks, with a further significant improvement in efficiency. The key advance on conventional PHY techniques is that both signalling waveforms and node signal processing is aware of the network topology and exploits it to improve overall network throughput.
The first part of the tutorial will introduce wireless network coding in the context of network information theory and the network-aware PHY layer. It will consider potential applications, from high-capacity next generation wireless broadband access networks to extremely power-efficient wireless sensor networks, pointing out the benefits of network awareness in such applications, and hence it will outline some potential scenarios in which it might operate. It will then introduce some of the basic principles and strategies of WNC, for example the strategies of hierarchical decode/compress and forward which replace the amplify/decode and forward strategies of conventional relaying, and consider the network capacity regions that result.
The second part will focus on an important example scenario: the two-source relay channel (2S-RC) in which two sources simultaneously exchange information with two destinations with the aid of a single relay, which may frequently occur in a wireless access network. It will describe code design for this scenario, and the effect of fading parameters on performance. In doing so it will introduce a fundamental principle which allows separation between an inner hierarchical exclusive alphabet and an outer error control code, which can be a conventional binary capacity-approaching code such as a turbo or LDPC code.
