A 6.72-Gb/s 8 pJ/bit/iteration IEEE 802.15.3c LDPC Decoder Chip

Zhixiang CHEN
Xiongxin ZHAO
Dajiang ZHOU
Satoshi GOTO

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E94-A    No.12    pp.2587-2596
Publication Date: 2011/12/01
Online ISSN: 1745-1337
DOI: 10.1587/transfun.E94.A.2587
Print ISSN: 0916-8508
Type of Manuscript: Special Section PAPER (Special Section on VLSI Design and CAD Algorithms)
Category: High-Level Synthesis and System-Level Design
WPAN,  IEEE802.15.3c,  LDPC decoder,  high data rate,  power-efficient,  

Full Text: PDF>>
Buy this Article

In this paper, we introduce an LDPC decoder design for decoding a length-672 multi-rate code adopted in IEEE 802.15.3c standard. The proposed decoder features high performances in both data rate and power efficiency. A macro-layer level fully parallel layered decoding architecture is proposed to support the throughput requirement in the standard. For the proposed decoder, it takes only 4 clock cycles to process one decoding iteration. While parallelism increases, the chip routing congestion problem becomes more severe because a more complicated interconnection network is needed for message passing during the decoding process. This problem is nicely solved by our proposed efficient message permutation scheme utilizing exploited parity check matrix features. The proposed message permutation network features high compatibility and zero-logic-gate VLSI implementation, which contribute to the remarkable improvements in both area utilization ratio and total gate count. Meanwhile, frame-level pipeline decoding is applied in the design to shorten the critical path. To verify the above techniques, the proposed decoder is implemented on a chip fabricated using Fujitsu 65 nm 1P12L LVT CMOS process. The chip occupies a core area of 1.30 mm2 with area utilization ratio 86.3%. According to the measurement results, working at 1.2 V, 400 MHz and 10 iterations the proposed decoder delivers a 6.72 Gb/s data throughput and dissipates a power of 537.6 mW, resulting in an energy efficiency 8.0 pJ/bit/iteration. Moreover, a decoder of the same architecture but with no pipeline stage for low-profile application is also implemented and evaluated at post-layout level.