干涉通道下有限區塊長度速率和最大化之混和單調性分析

Abstract

本文針對複數干擾信道下的有限長度正態近似的速率和優化問題。它使用干擾作為噪聲的解碼器來減少延遲，並取得高於傳統高斯碼本（球形高斯碼本）的 速率。它給出了兩種可實現的速率，分別是分開解碼（SD）和聯合解碼（JD）。 使用非凸函數優化工具——混合單調程序，本文在K個用戶的情況下提出了混合 單調函數。它提供了較低的SD速率和較低的JD速率，在多用戶情況下可以找到 不同混合單調函數的可實現速率。模擬結果表明，上述三個混合單調函數，其中 一個是用於較低JD速率，兩個是用於較低SD速率，在K個用戶的情況下可以取得非常相似的最大速率總和。本文還討論了不同可實現速率的不同混合單調性函數的迭代次數不同的問題。

This paper addresses the rate and optimization problems of the finite block length normal approximation under complex interference channels. It uses an interference-as-noise decoder to reduce the delay and achieve higher than the traditional Gaussian codebook— the spherical Gaussian codebook of the rate. It gives two achievable rate, the rate with separate decoding (SD) and the one with joint decoding (JD). And using the non-convex function optimization tool - mixed monotonic programming, this paper proposes mixed monotonic functions in the cases of K users for this optimization tool. It offers a lower SD rate and a lower JD rate, and the achievable rates can be found for different mixed monotonic functions in the case of multiple users. Simulations show that the above three mixed monotonic functions, one is for lower JD rate and two are for lower SD rate, can achieve very similar maximum rate sums in the case of K users. This paper also discusses the problem that the number of iterations of different mixed monotonicity functions of the proposed achievable rates differs.