Configurable and scalable high throughput turbo decoder architecture for multiple 4G wireless standards

  • Authors:
  • Yang Sun; Yuming Zhu;Manish Goel;Joseph R. Cavallaro

  • Affiliations:
  • ECE Department, Rice University. 6100 Main, Houston, TX 77005, USA;DSPS R&DCenter, Texas Instruments. 12500 TI Blvd MS 8649, Dallas, 75243, USA;DSPS R&DCenter, Texas Instruments. 12500 TI Blvd MS 8649, Dallas, 75243, USA;ECE Department, Rice University. 6100 Main, Houston, TX 77005, USA

  • Venue:
  • ASAP '08 Proceedings of the 2008 International Conference on Application-Specific Systems, Architectures and Processors
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we propose a novel multi-code turbo decoder architecture for 4G wireless systems. To support various 4G standards, a configurable multi-mode MAP (maximum a posteriori) decoder is designed for both binary and duo-binary turbo codes with small resource overhead (less than 10%) compared to the single-mode architecture. To achieve high data rates in 4G, we present a parallel turbo decoder architecture with scalable parallelism tailored to the given throughput requirements. High-level parallelism is achieved by employing contention-free interleavers. Multi-banked memory structure and routing network among memories and MAP decoders are designed to operate at full speed with parallel interleavers. We designed a very low-complexity recursive on-line address generator supporting multiple interleaving patterns, which avoids the interleaver address memory. Design trade-offs in terms of area and power efficiency are explored to find the optimal architectures. A 711 Mbps data rate is feasible with 32 Radix-4 MAP decoders running at 200 MHz clock rate.