Abstract
Provable lower bounds are presented for the information rate $I(X; X+S+N)$ where $X$ is the symbol drawn independently and uniformly from a finite-size alphabet, $S$ is a discrete-valued random variable (RV) and $N$ is a Gaussian RV. It is well known that with $S$ representing the precursor intersymbol interference (ISI) at the decision feedback equalizer (DFE) output, $I(X; X+S+N)$ serves as a tight lower bound for the symmetric information rate (SIR) as well as capacity of the ISI channel corrupted by Gaussian noise. When evaluated on a number of well-known finite-ISI channels, these new bounds provide a very similar level of tightness against the SIR to the conjectured lower bound by Shamai and Laroia at all signal-to-noise ratio (SNR) ranges, while being actually tighter when viewed closed up at high SNRs. The new lower bounds are obtained in two steps: First, a "mismatched" mutual information function is introduced which can be proved as a lower bound to $I(X; X+S+N)$. Secondly, this function is further bounded from below by an expression that can be computed easily via a few single-dimensional integrations with a small computational load.
Introduction
In digital communication, "capacity" defines the maximum reliable data transmission rate. For channels with memory, previous symbols interfere with the current one, known as Intersymbol Interference (ISI). Calculating the Symmetric Information Rate (SIR)—the capacity with uniform inputs—for ISI channels is computationally demanding. Traditional methods like Monte-Carlo simulations (BCJR algorithm) or high-dimensional integration (Hirt Bound) are often impractical for long-memory channels. Conversely, the popular Shamai-Laroia Conjecture (SLC) offers a fast approximation but lacks a formal proof. This paper introduces a provable lower bound that matches the SLC's efficiency and tightness while providing a solid mathematical foundation.
Methodology
The proposed approach utilizes two primary mathematical innovations:
- Mismatched Mutual Information (MMI): The authors define a new function, $I'(X; Y)$, which calculates entropy using a Gaussian distribution matching the interference variance rather than the true, complex probability density function. They rigorously prove this mismatched version serves as a lower bound to the true information rate.
- Cluster Identification: ISI interference is modeled as a "Gaussian mixture." To tighten the bound, the authors identify "dominant" interference taps (clusters). They calculate the exact distribution for these primary taps while treating the remaining "tail" of smaller taps as a simple variance term.
- Numerical Efficiency: By reducing the problem to a few single-dimensional integrals, the bound can be computed significantly faster than trellis-based simulations.
Experimental Setup
The bounds were evaluated using several standard synthetic ISI channel models, ranging from simple 2-tap channels to complex 10-tap models and magnetic recording simulations.
Key Results and Discussion
The new bound demonstrates remarkable tightness, often becoming indistinguishable from the SLC at low-to-mid SNRs. Notably, at high SNRs, the proposed bound can even exceed the SLC in tightness.
Key insights include:
- Provable Reliability: Unlike the SLC, this bound is mathematically proven, offering a guaranteed "floor" for the information rate.
- Dominant Tap Modeling: Modeling just the top 2 or 3 taps exactly provides a significant accuracy boost with minimal computational overhead.
- Computational Efficiency: The analytical bound achieves accuracy comparable to complex iterative simulations with a much lower computational load.
Limitations
While effective, the method's performance may degrade in "scattered" channels where no dominant taps exist. Additionally, the complexity scales exponentially with the number of chosen dominant clusters ($2^M$), and extending the method to very high-order alphabets would increase computational costs.
Reference: Jeong, S., & Moon, J. (2011). Easily Computed Lower Bounds on the Information Rate of Intersymbol Interference Channels. arXiv:1110.0560v1 [cs.IT].
Summary updated