Interference MAnaGement for Environmental-friendly NETworks – IMAGE-NET

In IMAGE-NET we will take a unifying view of interference and complexity, which will allow us to consider interference management solutions under strict delay and complexity constraints. Additionally, we will adopt a unifying view of the different methods of interference management, each defined by varying degrees of channel knowledge at the interfering nodes, as well as by varying capabilities of the different nodes. IMAGENET’s task is to combine these two perspectives.

IMAGE-NET approach will jointly provide both theoretical tools for analysis-and-optimization in wireless networks of interfering users, as well as provide clear breakthroughs towards computationally efficient implementation of the novel interference management methods.

IMAGE-NET will have both long term as well as short term impacts on society and economy.
In terms of short term goals it will address the economical and technical challenges of providing ultra-fast internet connectivity in urban and sub-urban areas. In terms of long term goals, IMAGE-NET seeks to be part of a paradigm shift in reducing the power requirements of telecommunication systems and their impact on the environment (electromagnetic radiation and carbon footprint).

Complete list at

imagenet.eurecom.fr/content/recent-publications

Achieving efficient wireless communications is widely believed to be pivotal in advancing important milestones in societies that value information and the environment. Such networks though, in the presence of a large number of users, can become an enormous environmental burden, given the massive amounts of power that is required both for transmitting signals as well as for supporting the intense computational process that take place. It has been well established that in such large networks, interference between nodes is a central bottleneck – which together with algorithmic complexity, take up the lion’s share of the power costs. The main volume of research in reducing user interferences has up to date focused on specifically exploring novel interference management techniques in settings that are simplistic, under very simplifying and extreme assumptions on the amount of channel knowledge at the nodes, and in the absence of considerations regarding implementation complexity which can often be entirely prohibitive.

There are two major aspects that clearly set the IMAGE-NET research proposal apart from the existing volume of work mentioned above. The first aspect is the fact that we will take a unified view of interference and complexity, where all the interference management solutions will be done under strict delay and complexity constraints. This approach is imperative given that a large fraction of the current state of art in interference management is far from implementable. The second aspect that sets this proposal apart is that we plan to adopt a unified view of the different methods of interference management, each defined by varying degrees of channel knowledge at the interfering nodes, as well as by varying capabilities of the different nodes. At the two extremes of this spectrum lie powerful but impractical interference alignment solutions that often require astronomical complexity, and on the other extreme lie very rare instances were simple linear solutions result in optimal interference management. Our task is to view these jointly. These differentiating aspects hold the promise of jointly providing both theoretical tools for analysis-and-optimization in wireless networks of interfering users, as well as providing clear breakthroughs towards computationally efficient implementation of the novel interference management methods.

To achieve this objective we identify three pertinent research areas related to the wireless networks. The study of fundamental tradeoff between interference management and feedback quality is proposed to evaluate the aspects related to tradeoff between the complexity of acquiring channel state information at transmitters (CSIT) and performance of interference management schemes, in particular the scenarios with no CSIT, reduced CSIT, and imperfect CSIT will be evaluated. We also look into the aspects related to the dissemination of CSIT in large networks to allow for network scalability. Thirdly we will also search for a unified view by studying the fundamental tradeoff between performance-delay-complexity in interference networks, and then proceed to propose algorithms that are simpler and under specific settings achieve optimal performance for interference management. Guided by this tradeoff we will seek to provide algorithms that are robust in the presence of imperfect CSIT and hybrid algorithms that combine some state of art interference management techniques to provide practical interference management solutions for networks.