Abstract
The weight generating functions associated with convolutional codes (CCs) are based on state space realizations or the weight adjacency matrices (WAMs). The MacWilliams identity for CCs on the WAMs was first established by Gluesing-Luerssen and Schneider in the case of minimal encoders, and generalized by Forney. We consider this problem in the viewpoint of constraint codes and obtain a simple and direct proof of this MacWilliams identity in the case of minimal encoders. For our purpose, we choose a different representation for the exact weight generating function (EWGF) of a block code, by defining it as a linear combination of orthonormal vectors in Dirac bra-ket notation. This representation provides great flexibility so that general split weight generating functions and their MacWilliams identities can be easily obtained from the MacWilliams identity for EWGFs. As a result, we also obtain the MacWilliams identity for the input-parity WAMs of a systematic convolutional code and its dual. Finally, paralleling the development of the classical case, we establish the MacWilliams identity for quantum CCs.
Introduction
In coding theory, the MacWilliams identity is a cornerstone for linear block codes, allowing researchers to calculate the error-correcting capabilities of a code by analyzing its dual. However, extending this identity to convolutional codes (CCs) has historically been difficult. Unlike block codes, CCs possess a memory-based structure typically described by state diagrams and Weight Adjacency Matrices (WAMs).
Previous attempts to establish a MacWilliams identity for CCs were often restricted to specific minimal encoders or relied on complex graph duality theorems that omitted explicit scaling factors. Furthermore, a clear extension to Quantum Convolutional Codes (QCCs) was missing. This paper provides a direct, transparent proof that unifies both classical and quantum cases.
Methodology
The authors employ two powerful conceptual tools to simplify the derivation:
Dirac Bra-Ket Notation
Borrowed from quantum mechanics, this notation represents the Exact Weight Generating Function (EWGF) as a linear combination of orthonormal vectors. This allows the MacWilliams identity to be viewed as a Fourier transform operation. This algebraic flexibility facilitates the handling of "split" weight functions, where different components of a codeword are treated independently.
Constraint Codes
Following an approach by G. David Forney, the authors analyze a convolutional code through its local behavior at each time step. By defining a block code that represents all valid transitions—including input, output, and state changes—they can apply block-code duality tools to the infinite-length structure of the convolutional code.
Key Insights
- Duality as a Fourier Transform: By representing weight enumerators in Hilbert space, the relationship between a code and its dual becomes a linear transformation. This unifies various types of weight enumerators (Hamming, input-parity, etc.) under a single mathematical framework.
- Explicit Scalar Factors: The authors explicitly calculate the scalar factor in the MacWilliams identity for CCs, addressing a detail often left as an implicit constant in previous literature.
- Quantum Parallelism: The paper demonstrates that Entanglement-Assisted Quantum Convolutional Codes (EA-QCCs) follow the same structural logic as classical CCs. By defining a quantum "constraint code," they establish the first MacWilliams identity for EA-QCCs.
- Systematic Encoders: The authors resolve an open question regarding the identity for "input-parity" weight adjacency matrices, which are critical for analyzing systematic codes used in modern applications like Turbo codes.
Results and Performance Analysis
The theoretical results are encapsulated in several key theorems:
- Theorems 7 & 10: Provide explicit MacWilliams identities for the WAM and the Full-trellis Input-Parity WAM (FIPWAM).
- Theorem 13: Establishes the MacWilliams identity for EA-QCCs, showing that the dual WAM is related to the original via a symplectic Fourier transform.
To validate these findings, the authors numerically verify the relationship between the Bit Error Rate (BER) of a code and its dual. As shown in the performance benchmarking, the BER of a convolutional code and its dual are nearly identical and are tightly upper-bounded by the derived weight enumerators.
Conclusion and Future Work
While this work simplifies the proof for minimal encoders, the assumption of minimality remains a common starting point. Future research may explore removing this constraint using more general graph duality. Additionally, these identities are essential for the error analysis of iterative decoding in Turbo codes. Applying these results to both classical and quantum Turbo codes represents a promising direction for ongoing investigation, particularly for complex or non-binary channels where the EWGF approach is most advantageous.
Summary updated