Killing two birds with one bit of feedback: performance/complexity ramifications of feedback - new results and open problems

Arun Kumar SINGH - Post-Doc, CM department
Communication systems

Date: -
Location: Eurecom

Outage limited communications are known to yield reduced reliability, and to also demand prohibitively large (exponential) computational resources. What if though we were offered a single bit of feedback? What could we achieve? In this talk we try to answer the above by focusing on two fundamental questions. The first question asks what is the complexity savings that feedback provides for a given fixed rate-reliability performance - in other words, what is the best way to use feedback to reduce complexity without worrying about improving performance. The second question is much harder, and it basically asks what is the complexity costs of achieving the full rate-reliability benefits of feedback - in other words, how many flops do we really need to convert outage-limited wireless communications to ergodic-like communications. We discuss some of our surprising findings, while also offering a glimpse to the main unresolved questions.