How to best process and forward received signals, or relay, in wireless networks has been an open question for many years. We will revisit determining the capacity of relay networks by taking a fresh look at the problem focussed on understanding the potential of exploiting two types of codebook structure in relay networks: 1) messages received at relays, even those received at rates above capacity, originated from codebooks. This codebook structure should be harnessed in relaying and we seek to determine how to best compress signals coded at rates above capacity; 2) matching codebook structure (e.g. linearity of lattice codes is well suited to linear additive Gaussian noise channels) to channel structure has recently been shown to be efficient in decoding functions of messages. In this thrust, we seek to make precise how and when structured codes may be exploited in multi-source relay networks. Our two approaches both involve novel departures from and additions to our understanding of random coding based schemes.
A full solution would constitute a significant step towards the understanding of network information theory, and will provide new insight into the elusive relay channel. However, the significance of solving this problem goes well beyond information theory: understanding how to most efficiently relay in wireless networks is of direct relevance to satiating the demands on WiFi, cellular, ad hoc military and first responder wireless networks of the present and future. The fundamental limit -- or capacity -- of such networks not only acts as a benchmark for engineers building systems, but also provides guidance as to what schemes perform well. The PI will integrate portions of this research into UIC's curriculum and will integrate this theoretical approach with practice through the PI-founded UIC Software Defined Radio lab.
