[Paper Review] Distributed Joint Source-Channel Coding for arbitrary memoryless correlated sources and Source coding for Markov correlated sources using LDPC codes

Vaneet Aggarwal|ArXiv.org|Feb 19, 2008

Wireless Communication Security Techniques22 references3 citations

TL;DR

This paper proposes a distributed joint source-channel coding scheme using LDPC codes for arbitrary memoryless correlated sources and Markov-correlated sources across any point in the Slepian-Wolf rate region. It extends Wyner's syndrome concept by applying belief propagation over LDPC codes with tailored message-passing rules, achieving near-capacity performance for both joint source-channel coding and distributed source coding under arbitrary correlation models and channel conditions.

ABSTRACT

In this paper, we give a distributed joint source channel coding scheme for arbitrary correlated sources for arbitrary point in the Slepian-Wolf rate region, and arbitrary link capacities using LDPC codes. We consider the Slepian-Wolf setting of two sources and one destination, with one of the sources derived from the other source by some correlation model known at the decoder. Distributed encoding and separate decoding is used for the two sources. We also give a distributed source coding scheme when the source correlation has memory to achieve any point in the Slepian-Wolf rate achievable region. In this setting, we perform separate encoding but joint decoding.

Motivation & Objective

• To develop a practical distributed joint source-channel coding scheme for arbitrary memoryless correlated sources at any point in the Slepian-Wolf rate region.
• To extend the use of LDPC codes beyond binary symmetric channel models to arbitrary correlation structures and arbitrary link capacities.
• To address distributed source coding for Markov-correlated sources using LDPC codes with joint decoding and belief propagation.
• To demonstrate near-capacity performance through simulations using irregular LDPC codes and iterative decoding.
• To bridge the gap between theoretical Slepian-Wolf limits and practical implementation using LDPC-based constructions.

Proposed method

• Adapts Wyner's syndrome approach by modeling source correlation as a binary channel and using LDPC codes for encoding, with systematic bits and parity bits distributed across sources.
• Employs belief propagation over a Markov model to compute a priori probabilities for state transitions, using likelihoods derived from the correlation model.
• Uses message-passing between bit nodes, check nodes, and the Markov model, with iterative updates based on reliability metrics (|0.5 - a_n^x| vs |0.5 - b_n^x|) to prioritize updates.
• Applies unequal error protection by assigning higher variable node degrees to information bits and lower degrees to parity bits in the LDPC code design.
• Implements joint decoding via iterative belief propagation: messages are passed from bit nodes to the Markov model, then to check nodes, and back to bit nodes until convergence.
• Uses normalizing factors (η_n) to ensure posterior probabilities sum to one during message updates, maintaining consistency in belief propagation.

Experimental results

Research questions

• RQ1 Can LDPC codes be effectively used to achieve any point in the Slepian-Wolf rate region for arbitrary memoryless correlated sources?
• RQ2 How can joint source-channel coding be designed for arbitrary source correlation models and arbitrary channel capacities using LDPC codes?
• RQ3 Can belief propagation over Markov models improve decoding performance in distributed source coding with memory-structured correlations?
• RQ4 What is the impact of irregular LDPC code design on performance in distributed source coding with Markov sources?
• RQ5 How does unequal error protection in LDPC codes affect decoding reliability in joint source-channel coding schemes?

Key findings

• The proposed scheme achieves near-capacity performance for arbitrary memoryless correlated sources across the entire Slepian-Wolf rate region using LDPC codes.
• Simulations show that with k=6250, α=0.2, and R_X1=R_X2=0.75, the system achieves a rate of 0.8421 with C_f=C_b=0.5 under Markov correlation.
• The use of irregular LDPC codes (e.g., (3,x) with tailored degree distributions) improves performance over regular codes, especially in high-SNR regimes.
• Belief propagation with state-dependent likelihoods and iterative message-passing enables reliable decoding even when only partial information is available at each encoder.
• The scheme successfully decodes both sources using only the received parity and systematic bits, with no inter-source communication during encoding.
• The method generalizes prior work limited to symmetric or asymmetric points in the Slepian-Wolf region, enabling arbitrary rate allocation.