580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can state-space representations optimize the construction and decoding of product convolutional codes?
This research area investigates the algebraic and system-theoretical frameworks for product convolutional codes using minimal state-space representations. The motivation is to achieve systematic constructions of product codes that are minimal, reachable, and observable, facilitating efficient encoding and decoding operations. The state-space approach provides a foundation to extend known results from block product codes to convolutional codes, potentially enhancing error correction and burst error handling.
Key finding: Demonstrates a constructive methodology to derive minimal state-space representations of product convolutional codes from two minimal state-space representations of the horizontal and vertical component convolutional codes.... Read more
Key finding: Introduces an algorithm to obtain equivalent encoding matrices for convolutional codes that maximize the distances of partitioned subcodes via path puncturing. This achieves enhanced partitionings for generalized... Read more
Key finding: Develops a framework viewing serial and parallel concatenated convolutional codes as special cases of partially concatenated convolutional codes, with an intermediary partitioning matrix controlling the inner encoder input.... Read more
2. What design principles enable high-performance serial and parallel concatenated convolutional codes with optimized error correction?
This theme covers the analysis, design, and performance optimization of serially and parallelly concatenated convolutional codes (SCCCs and PCCCs), including their rate-compatible and turbo decoding variants. The focus lies on the choice of constituent convolutional codes, interleaver design, puncturing patterns (including inner code systematic and parity bits), and their impact on free distance, error floors, and decoding thresholds. Understanding the interplay between these parameters is crucial for approaching Shannon capacity limits and tailoring codes for various communication channel conditions.
Key finding: Systematically evaluates SCCCs based on different constituent recursive systematic convolutional codes and interleaver lengths, establishing that the overall coding gain derives from effective free distances determined by... Read more
Key finding: Establishes design guidelines for parallel concatenated convolutional codes based on systematic recursive constituent codes with interleavers, emphasizing the crucial role of effective free distance in performance. Shows that... Read more
Key finding: Performs analytical upper bounds and error probability analysis for a novel class of SCCCs allowing inner encoder puncturing beyond unit rate, including both systematic and parity bits. Determines that optimal inner... Read more
Key finding: Introduces rate-compatible SCCCs with puncturing applied to both inner systematic and parity bits, enabling code rates beyond the outer encoder rate. Demonstrates that the puncturing strategy must be adapted to the target SNR... Read more
Key finding: Investigates spatially coupled SCCCs (SC-SCCs) demonstrating that increasing coupling memory improves decoding thresholds and error floors, but at increased latency and complexity. Proposes design criteria enabling flexible... Read more
3. How can low-density parity-check codes and convolutional code constraints be integrated and optimized for enhanced decoding thresholds and complexity tradeoffs?
This line of research explores generalized LDPC codes with convolutional code component constraints (CC-GLDPC), their irregularity, threshold improvements, encoding complexity, and their comparison to classical LDPC codes. Emphasis is placed on designing codes for efficient belief propagation decoding with strong component codes represented as trellis constraints, enabling spatial coupling and tailored performance-to-complexity ratios in modern communication systems such as 5G. The research also includes FPGA performance comparisons of turbo and LDPC codes under different hardware constraints.
Key finding: Implements a turbo coding scheme based on serially concatenated convolutional encoders combined with trellis-coded modulation (TCM) for Rician fading channels. Demonstrates through MATLAB simulations a 5 dB performance gain... Read more
Key finding: Develops irregular CC-GLDPC code ensembles with constraint nodes formed by convolutional codes of varying memories and rates, enhancing belief propagation (BP) decoding thresholds significantly over regular ensembles with... Read more
Key finding: Introduces a family of (d_v,d_c)-regular GLDPC codes with convolutional code constraints extending braided convolutional codes to arbitrary regular graphs. Conducts density evolution and finite-length weight enumerator... Read more
Key finding: Presents algebraic constructions of regular LDPC codes derived from combinatorial designs such as Steiner 2-designs and partial geometries, guaranteeing desirable code properties including girth and stopping set distribution.... Read more
Key finding: Provides comparative FPGA hardware implementation and performance analyses of LDPC and turbo codes for wireless communication. Designs encoder and decoder architectures for both coding schemes on Virtex-5 FPGA and evaluates... Read more
Key finding: Proposes a variant of the Offset Min-Sum decoding algorithm optimized for 5G NR LDPC codes in partially parallel layered architectures. By approximating the secondary minimum with a variable weight added to the primary... Read more