[Paper Review] Error Exponent for Multiple-Access Channels:Lower Bounds

Ali Nazari, Achilleas Anastasopoulos+1 more|arXiv (Cornell University)|Oct 7, 2010

Wireless Communication Security Techniques14 references9 citations

TL;DR

This paper develops three new lower bounds on the error exponent for two-user discrete memoryless multiple-access channels (DM-MACs): random coding, typical random coding, and expurgated bounds. Using a constant composition code ensemble with α-decoding, it proves that the typical random coding and expurgated bounds strictly outperform the standard random coding bound at low rates, with the first explicit use of expurgation for MACs.

ABSTRACT

A unified framework to obtain all known lower bounds (random coding, typical random coding and expurgated bound) on the reliability function of a point-to-point discrete memoryless channel (DMC) is presented. By using a similar idea for a two-user discrete memoryless (DM) multiple-access channel (MAC), three lower bounds on the reliability function are derived. The first one (random coding) is identical to the best known lower bound on the reliability function of DM-MAC. It is shown that the random coding bound is the performance of the average code in the constant composition code ensemble. The second bound (Typical random coding) is the typical performance of the constant composition code ensemble. To derive the third bound (expurgated), we eliminate some of the codewords from the codebook with larger rate. This is the first bound of this type that explicitly uses the method of expurgation for MACs. It is shown that the exponent of the typical random coding and the expurgated bounds are greater than or equal to the exponent of the known random coding bounds for all rate pairs. Moreover, an example is given where the exponent of the expurgated bound is strictly larger. All these bounds can be universally obtained for all discrete memoryless MACs with given input and output alphabets.

Motivation & Objective

• To extend the framework of error exponent analysis from point-to-point channels to multiple-access channels (MACs).
• To address the lack of advanced error exponent bounds for discrete memoryless MACs compared to point-to-point channels.
• To develop new lower bounds on the reliability function of DM-MACs that outperform existing bounds.
• To introduce and formalize the method of expurgation for MACs, which had not been previously applied due to complexity in handling correlated codeword pairs.
• To provide universally applicable bounds across all discrete memoryless MACs with given input and output alphabets.

Proposed method

• Adapts the α-decoding framework and constant composition code ensemble from point-to-point channels to the two-user MAC setting.
• Derives the random coding bound as the average error exponent over the constant composition code ensemble.
• Introduces the typical random coding bound as the performance of typical codes in the ensemble, showing it exceeds the random coding exponent.
• Proposes an expurgated bound by removing high-rate codewords from the codebook, improving performance at low rates.
• Uses information-theoretic optimization over joint types (V) to characterize the error exponent bounds, with constraints on mutual information and entropy terms.
• Employs continuity arguments and type class counting to bound the average error probability and derive exponential decay rates.

Experimental results

Research questions

• RQ1 Can the random coding, typical random coding, and expurgated bounds—well-established for point-to-point channels—be generalized to multiple-access channels?
• RQ2 Does the typical random coding bound for MACs outperform the standard random coding bound in terms of error exponent?
• RQ3 Can the method of expurgation be meaningfully applied to MACs, and does it yield a strictly better error exponent than existing bounds?
• RQ4 Are the new bounds universally applicable across all discrete memoryless MACs with fixed input and output alphabets?
• RQ5 What is the relationship between the exponents of the three new bounds, and in which rate regions do they dominate?

Key findings

• The typical random coding bound has a strictly larger error exponent than the standard random coding bound for all rate pairs.
• The expurgated bound has a strictly larger error exponent than the random coding bound in at least one example, demonstrating a performance gain at low rates.
• The expurgated bound is the first explicit application of expurgation to MACs, overcoming the challenge of correlated input sequences after codeword elimination.
• All three bounds—random coding, typical random coding, and expurgated—are universally applicable to all discrete memoryless MACs with given input and output alphabets.
• The bounds are derived using a constant composition code ensemble with α-decoding, enabling tighter analysis of typical codebook performance.
• The error exponent expressions are derived via optimization over joint types (V), with constraints ensuring rate and mutual information conditions are met.