Abstract:The channel polarization under the channel noise with memory is comprehensively studied. With the help of the genie-aided channel, we prove that the polarized channels also converge to extremal channels under the standard polar codes structure. More importantly, the ratio of the perfect channel can be larger than $I(W)$ which is the capacity of the original channel. However, the polarization rate is shown to be slower than the binary-input discrete memoryless channel (DMC) case. Specifically, the upper bound of the block error is $\mathcal{O}(L^{-c_0})$ where $L$ is the block length and $c_0$ is some positive constant. Furthermore, the upper and lower bound of the gap between the capacity and cutoff rate is investigated when the block length is finite, which is more useful for practical applications.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: the study of the channel polarization phenomenon under the condition of channel noise with memory. Specifically, the authors explore whether the polarized channels under the standard polar code structure will still converge to extreme channels when the channel noise has memory, and analyze the polarization rate and performance gain.
### Main problems and objectives
1. **Convergence of polarized channels**:
- The authors prove that with the help of the genie - aided channel, even if the channel noise has memory, the polarized channels will still converge to extreme channels.
2. **Proportion of perfect channels**:
- The research finds that after full polarization, the proportion of perfect sub - channels can exceed the capacity \(I(W)\) of the original channel, that is, reach \(I(W)+c\), where \(c\) is a non - negative constant. This result is different from the situation in traditional memoryless noise channels.
3. **Polarization rate**:
- The paper points out that the polarization rate of channels with memory noise is slower than that of binary - input discrete memoryless channels (DMC). Specifically, the upper bound of the block error rate is \(O(L^{-c_0})\), where \(L\) is the block length and \(c_0\) is a certain positive constant.
4. **Gap between cut - off rate and capacity**:
- The authors also study the upper and lower bounds of the gap between the cut - off rate and capacity under a finite block length. This result helps to quantify the impact due to incomplete polarization.
### Methods and techniques
To overcome the challenges brought by memory noise, the authors introduce the genie - aided channel, assuming that all previous noise samples are known. In this way, they can decompose the mutual information of the composite channels and derive relevant mathematical formulas. In addition, the paper also uses tools such as Markov processes and Bhattacharyya parameters to analyze the channel polarization process.
### Main conclusions
1. **Increase in total mutual information**:
- After complete polarization, the total mutual information increases compared to the original channel, specifically \(I(W)+I^{\dagger}(W)\), where \(I^{\dagger}(W)=I(N_1;N_0)-\frac{1}{2}I(Y_1;Y_0)-I(Y_0;Y_{1}^{+\infty})\geq0\).
2. **Slower polarization rate**:
- Compared with memoryless noise channels, the polarization rate of channels with memory noise is slower, and the upper bound of the block error rate is \(O(L^{-c_0})\).
3. **Gap between cut - off rate and capacity**:
- Provides the upper and lower bounds of the gap between the cut - off rate and capacity under a finite block length, which is more useful in practical applications.
Through these studies, the paper not only expands the application scope of polar code theory but also provides new insights and methods for dealing with communication systems with memory noise.