Turbo Decoder Design based on an LUT-Normalized Log-MAP Algorithm

Area of Science:

• Wireless communication systems
• Digital signal processing
• Error correction coding

Background:

• Turbo codes are crucial for error correction in wireless communications.
• Existing approximations like Log-MAP (Log maximum a posteriori) and Max-Log-MAP in TD-LTE face challenges with high complexity or bit error rates.
• Efficient decoding algorithms are needed to balance performance and resource utilization.

Purpose of the Study:

• To propose a novel normalized Log-MAP (Nor-Log-MAP) decoding algorithm.
• To develop a hybrid LUT-Nor-Log-MAP algorithm by integrating Nor-Log-MAP with LUT-Log-MAP.
• To design a dedicated Normalization Functional Unit (NFU) for Soft-Input Soft-Output (SISO) decoder computing units.

Main Methods:

• Approximating the max* function using a fixed normalized factor multiplied by the max function within the Log-MAP algorithm.
• Combining the Nor-Log-MAP algorithm with a Lookup-Table Log-MAP (LUT-Log-MAP) algorithm.
• Implementing a Normalization Functional Unit (NFU) for SISO decoders.

Main Results:

• The LUT-Nor-Log-MAP algorithm achieves decoding performance close to the LUT-Log-MAP algorithm.
• The LUT-Nor-Log-MAP algorithm demonstrates a resource saving of approximately 2.1% compared to the LUT-Log-MAP algorithm.
• Compared to the Max-Log-MAP algorithm, LUT-Nor-Log-MAP shows a 0.25-0.5 dB gain in decoding performance.
• A Turbo decoder utilizing the Nor-Log-MAP approach achieved 36 Mbit/s throughput on a Cyclone IV platform at 44 MHz.

Conclusions:

• The proposed Nor-Log-MAP algorithm and its LUT-Nor-Log-MAP variant offer an effective trade-off between decoding performance and hardware complexity.
• The developed NFU contributes to efficient SISO decoder design.
• The results indicate significant improvements in resource efficiency and error correction capabilities for Turbo decoders in wireless communication standards like TD-LTE.