This project deals with models of wireless multi-terminal networks incorporating practical constraints such as individual links that experience fading, applications that are delay-sensitive, network communication that is subject to broadcast and interference constraints and nodes that are constrained to operate in half-duplex mode. The network is assumed to be static for the duration of the message, but can change from one message to the next and channel-state information is assumed to be present only at the receiver. In such settings, cooperative communication in which intermediate nodes facilitate communication between a particular source-sink pair, is key to efficient operation of the network.
A key goal of any communication system, is one of achieving an optimal rate-reliability tradeoff. The diversity-multiplexing gain tradeoff (DMT) determines the tradeoff between relevant first-order approximations to the rate and reliability of communication. The DMT of point-to-point communication links has been extensively studied and signal sets are available that are optimal under any statistical distribution of the fading channel. There now exist protocols and codes for two-hop relay networks that come close to achieving the corresponding min-cut upper bound on DMT. Goals of this project include: 1) determining the DMT of various classes of multiterminal networks ranging from broadcast, cooperative-broadcast and multiple-access channel networks to layered multi-hop networks; 2) identifying the classes of networks for which the DMT of the network is given by the DMT of the min-cut; 3) assessing the impact of asynchronous operation of the network, as well as of the presence of feedback along one or more links in the network; 4) the construction of codes with lesser decoding complexity.
This research project has two goals. The first is developing innovative coding techniques for wireless networks. The second is developing practical techniques in interference alignment, which allows multiple wireless users to communicate using the same frequency channel. The first goal aims at improving network performance by minimizing the impact of channel outage (such as fading in mobile channels and blockage in directional links). Novel packet code schemes have been constructed, which manages retransmission in the event of data loss without impeding efficiency. Such codes can be used for point-to-point links, relayed links, and networks. Analytical and simulation works have shown advantage of the new techniques over existing ones. Towards the same end, this research also developed a mathematical tool to solve the problem of optimally configuring the radios in a network in terms of resource allocation and multi-antenna operation modes, to maximize network efficiency. This tool is being translated into a software package. The second goal seeks to bring the interference alignment technique closer to practical use. Invented in 2008, interference alignment is a technique that can potentially allow many users to use the same channel while losing only half of the data rate per user. Therefore, the sum data rate carried by the network is greatly increased. It achieves such benefit by designing every user’s signal so that all mutual interference is "compressed", or aligned, into a limited vector subspace, leaving the rest of the vector space for interference-free communications. However, there is a theory predicting an upper bound on the performance improvement under constraints. According to this upper bound, a practical system (i.e., a system with limited delay and bandwidth) can achieve only a small gain by using interference alignment. However, there are reasons to believe that such an upper bound can be exceeded, as least in some special cases pertaining to practical systems such as the 4G cellular and WiFi. Therefore, it is believed that the upper bound was based on assumptions that are too restrictive. This research attempts to understand the mathematical reason behind the exceedance of this predicted upper bound, and aims to find better ways to realize interference alignment gains under practical constraints. Mathematical evidences leading to the upper bound were examined carefully and were found insufficient to prove the case in general. Some counter examples have been constructed to better understand the issue. Numerical tools have been constructed to provide empirical clues that may help infinding an alternative theory. Although not reaching full maturity during the course of this project, the work conducted in this project broke new ground in the field of wireless communications that may lead significant gains in performance. With the advent of new consumer products and services such as smart phones, wearable devices, and cloud computing, the demand for high speed and reliable wireless communications has exploded in this decade. Innovations and techniques that can improve wireless network performances may provide a positive impact to the economies and quality of life. This project also contributes to academic advances. It has produced two peer-reviewed conference reports, one submitted conference paper and one submitted journal paper, with two more papers in preparation. One graduate student was supported by the grant and two additional graduate students worked in the project under the mentorship of the principal investigators. Collaborations with researches in other institutes are also part of the research effort. The investigators plan to continue this research beyond the funded NSF effort.
