Improving the Reliability of Next-Generation Wireless Communication Systems
Korean researchers develop an algorithm that improves the speed at which error-prone structures in Raptor-like LDPC codes are identified
Raptor-like low-density parity-check (LDPC) codes are used to correct errors in transmitted messages in communication systems that require high speeds and low latencies. However, the decoding process is hindered by the presence of trapping sets. Now, researchers have developed an algorithm that can identify trapping sets more quickly, in a fraction of the time that existing algorithms require. The algorithm can lead to more reliable communication systems for applications like factory automation and power distribution.
Ultra-reliable low-latency communication systems, like 5G, allow devices to communicate and exchange information at high speeds. Such a connected ecosystem can support a variety of applications such as wireless factory automation, eHealth, and smart power distribution. To ensure the reliability of the transmitted messages, such communication systems must have an error-correcting process that can quickly correct errors. Raptor-like low-density parity-check (LDPC) codes are well suited for these kinds of applications where low error rates are required.
Transmitted messages contain several binary digits (0s and 1s) in a sequence. Based on the values of the bits in the message, the LDPC code adds extra bits known as (parity bits) to the original message to form a codeword that is sent to the receiver. At the receiver, a decoder checks the received codeword for errors. If errors are detected, it uses an iterative process to correct the errors and reconstruct the original message. This process involves updating the values of the original message, based on the constraints imposed by the parity bits. However, error correction fails whenever absorbing sets are encountered during the decoding process.
Identifying these sets can be time-consuming and computationally intensive, making it challenging to design decoding algorithms that can improve the error correction performance of LDPC codes.
Now, in a study published in Electronic and Telecommunications Research Institute (ETRI) Journal, researchers from Electronics and Telecommunications Research Institute, Korea, have developed an algorithm that can identify dominant absorbing sets faster than existing search algorithms.
“The proposed algorithm is essential for a design of short-block length LDPC code and related rate matching techniques and can be further used for research on iterative detection and decoder or deep learning-based receiver design,” says Mr. Woncheol Cho, the corresponding author of the study.
LDPC codes use a parity-check matrix to generate a codeword which is decoded using a Tanner graph that represents the parity-check matrix. The graph consists of variable nodes, which represent the bits in the codeword, and check nodes, which represent constraints on the variable nodes. To reconstruct the transmitted message, a belief propagation algorithm iteratively updates the values of the variable nodes until it finds values that satisfy the constraints imposed by the check nodes. However, certain nodes can be incorrectly connected, resulting in the decoding algorithm converging on an incorrect value. Such subsets of incorrect nodes are known as trapping sets, among which, fully absorbing sets (FASs) are a common cause of the error-rate floor for LDPC codes. To detect these absorbing sets faster, the proposed algorithm limits the search space to a subgraph of the Tanner graph that contains most of the absorbing sets.
“The proposed algorithm finds more dominant fully absorbing sets within one-seventh of the computation time of the existing search algorithm,” says Mr. Cho.
The proposed method can be widely used to lower the error rate of LDPC codes, making it suitable for current high-speed communication systems and next-generation wireless technologies that will be crucial in setting up the Internet of Things (IoT) services and an interconnected, data-driven network.
Reference
Titles of original papers: Effective identification of dominant fully absorbing sets for Raptor-like LDPC codes
Name of author: Woncheol Cho, Chanho Yoon, Kapseok Chang, Young-Jo Ko
Affiliation: Electronics and Telecommunications Research Institute, Republic of Korea
About Woncheol Cho
Woncheol Cho is a Research Engineer at the Mobile Communication Research Division, Electronics and Telecommunications Research Institute (ETRI), Daejeon, Republic of Korea since 2020. His research group has worked on absolute time synchronization, diversity-joint channel coding, new waveform design, and multi-antenna technologies for beyond-5G/6G high-capacity ultra-precision communication systems. His research interests include channel coding, coded modulation, and mmWave beamforming. He received his BS degree at the School of Undergraduate Studies and MS degree at EECS,Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Republic of Korea, in 2018 and 2020, respectively. He can be reached at woncheol@etri.re.kr.