An open problem in polarization theory is to determine the bi nary operations that always lead to polarization when they are used in Arıkan-like constructions. This paper , which is presented in two parts, solves this problem by providing a necessary and sufficient condition for a binar y operation to be polarizing. This (first) part of the paper introduces the mathematical framework that we will us e in the second part [1] to characterize the polarizing operations: uniformity-preserving, irreducible, ergodi c and strongly ergodic operations are defined. The concepts of a stable partition and the residue of a stable partition ar e introduced. We show that an ergodic operation is strongly ergodic if and only if all its stable partitions are their own residues. We also study the products of binary operations and the structure of their stable partitions. We show that the product of a sequence of binary operations is strongly ergodic if and only if all the operations in the se qu nce are strongly ergodic. In the second part of the paper, we provide a foundation of polarization theory based on the ergodic theory of binary operations that we develop in this part.
[1]
Rajai Nasser,et al.
Polarization theorems for arbitrary DMCs
,
2013,
2013 IEEE International Symposium on Information Theory.
[2]
Aria Ghasemian Sahebi,et al.
Multilevel polarization of polar codes over arbitrary discrete memoryless channels
,
2011,
2011 49th Annual Allerton Conference on Communication, Control, and Computing (Allerton).
[3]
Eren Sasoglu.
Polar codes for discrete alphabets
,
2012,
2012 IEEE International Symposium on Information Theory Proceedings.
[4]
Emre Telatar,et al.
Polarization for arbitrary discrete memoryless channels
,
2009,
2009 IEEE Information Theory Workshop.
[5]
Alexander Barg,et al.
Polar Codes for-Ary Channels ,
,
2013
.
[6]
Erdal Arikan,et al.
Channel Polarization: A Method for Constructing Capacity-Achieving Codes for Symmetric Binary-Input Memoryless Channels
,
2008,
IEEE Transactions on Information Theory.