A systematic study of Physical LAyer Network coding: from information-theoretic understanding to practical DSP algorithm design (P.L.A.N)

High spectral efficiency is the holy grail of wireless networks due to the well-known scarcity of radio spectrum. While up to recently there seemed to be no way out of the apparent end of the road in spectral efficiency growth, the emerging approach of Network Coding has cast new light in the spectral efficiency prospects of wireless networks [1]. Initial results have demonstrated that the use of network coding increases the spectral efficiency up to 50% [2, 3]. Such a significant performance gain is crucial for many important bandwidth-hungry applications such as broadband cellular systems, wireless sensor networks, underwater communication scenarios, etc. Currently network coding has received a lot of attention from the wireless communication community; however, many existing works focused on the application of network coding to upper layers and the study of its impact on the physical layer (PHY) design only began recently. The aim of this proposal is to systematically study network coding at the physical layer, where we will not only characterize the fundamental limits of physical layer network coding, but also design practical digital signal processing (DSP) algorithms to realize the performance gain promised by those theoretic results. The novelty of the proposed project lies on the fact that this project will be the first UK effort to bridge information-theoretic studies and DSP algorithm design for PHY network coding. This will be done by first deriving the capacity region of network coding, which provides us the upper bound of the system performance. With such a better understanding, we will develop efficient transmission protocols and DSP algorithms to realize such optimal performance in practice. Interference alignment, a technology recently developed to cope with co-channel interference, will be applied to network coding transmissions for further performance improvement. Information-theoretic results, such as outage and symbol error probabilities, will be developed and testbed-based experimental evaluation will be carried out, so a more insightful understanding for our developed schemes can be obtained.

This work is proposed at a time when network coding has been envisioned to bring the fundamental changes to the way communication systems are designed, operated and understood. Significant performance gain promised by network coding is important to broadband mobile communications, where both the uplink and downlink data rate can be improved dramatically. Other applications of network coding include real-time sensor networks which can be used for environmental monitoring of physical and biological indicators, tactical surveillance, disaster prevention, undersea exploration, assisted navigation, etc. However, compared to the significant progress in the application of network coding to the upper layer design, there is less progress reported to date on studying network coding at the physical layer, especially in the UK. The novelty of this project is four-fold: firstly characterize the data rates achieved by network coding for various multi-user scenarios; secondly devise spectrally efficient transmission protocols by utilizing the concept of interference alignment and efficiently combating co-channel interference; thirdly design new low-complexity mapping and receive DSP algorithms to realize the performance gain of network coding in practical systems; fourthly carry out experimental evaluations to investigate how the proposed algorithms perform in practice in conjunction with leading technology providers and end-users. The proposal systematically addresses the fundamental issues of network coding from the information-theoretic aspect to DSP-enabled communications, which is beneficial for the communications community to get a better understanding of network coding. Also such obtained insights provide a precise guideline for the efficient design of practical and reliable wireless systems. Following this, the outcomes of this research will also be of considerable value to UK companies involved in developing broadband communication systems or real-time wireless sensor networks for the purpose of infrastructure monitoring or military controlling, as well as robust vehicular networks to support diverse quality of service. Indeed, the potential of the project cannot be overstated and this is evidenced by strong industrial support from Bell-Lab, BP and Infineon.