Jessalyn Bolkema – University of Nebraska-Lincoln

The tensor-like join: graph-theoretic insights into polar code performance

Arikan’s polar codes have been celebrated for their capacity-achieving performance since first presented in 2008; however, the question of optimal finite-length design and decoding remains open. To understand the efficacy of graph-based decoding algorithms in this context, we have introduced the tensor-like join: a graph operation for constructing realizations of polar codes. In this talk, we will present combinatoric and graph-theoretic results on decoding impediments and explore the role of the tensor-like join in reaching these conclusions.