MLMin Li

Min Li

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Min Li has a total of 21 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Exploration of Soft-Output MIMO Detector Implementations on Massive Parallel Processors

    by Robert Fasthuber, Min Li, David Novo, Praveen Raghavan, Liesbet Van Der Perre, Francky Catthoor


  • Energy Aware Signal Processing for Software Defined Radio Baseband Implementation

    by Min Li, David Novo, Bruno Bougard, Claude Desset, Antoine Dejonghe, Liesbet Van Der Perre, Francky Catthoor


  • Optimizing Near-ML MIMO Detector for SDR Baseband on Parallel Programmable Architectures

    by Min Li, Bruno Bougard, Weiyu Xu, David Novo, Liesbet Van Der Perre, Francky Catthoor

    ML and near-ML MIMO detectors have attracted a lot of interest in recent years. However, almost all the reported implementations are delivered in ASICs or FPGAs. Our contribution is optimizing the near-ML MIMO detector for parallel programmable architectures, such as those with ILP and DLP features. In the proposed SSFE (selective spanning with fast enumeration), architecture-friendliness is explicitly introduced from the very beginning of the design flow. Importantly, high level algorithmic transformations make the dataflow pattern and structure fit architecture-characteristics very well. We enable abundant vector-parallelism with highly regular and deterministic dataflow in the SSFE; memory rearrangements, shuffling and non-predictable dynamism are all elaborately excluded. Hence, the SSFE can be easily parallelized and efficiently mapped onto ILP and DLP architectures. Furthermore, to fine-tune the SSFE on parallel architectures, extensive pre-compiler transformations are applied with the help of the application-level information. These optimize not only computation-operations but also address-generations and memory-accesses. Experiments show that the SSFE brings very efficient resource-utilizations on real-life VLIW architectures. Specifically, with the SSFE the percentage of NOPs instructions on VLIW is below 1%, even better than that achieved by the software-pipelined FFT. To the best of our knowledge, this is the first reported work about comprehensive optimizations of near-ML MIMO detectors for parallel programmable architectures.


  • Generic Multi-Phase Software-Pipelined Partial-FFT on Instruction-Level-Parallel Architectures and SDR Baseband Applications

    by Min Li, David Novo, Bruno Bougard, Liesbet Van Der Perre, Francky Catthoor

    The PFFT (Partial FFT) is an extended FFT where only part of input or output bins are used. By pruning the useless dataflow, the PFFT can potentially achieve a significant speedup in many important applications. Although theoretical aspects of the PFFT have been thoroughly studied in past three decades, efficient implementations were rarely reported. The most important obstacle is the highly irregular dataflow and the associated control flow. In addition, a size-N PFFT has 2 N dataflow possibilities, so that delivering both flexibility and efficiency in the same implementation is very challenging. This paper presents a generic scheme to map the highly irregular dataflow of arbitrary PFFT onto ILP architectures with highly efficient SWP (Software-Pipelining). Constraints and opportunities of algorithms and architecture are carefully analyzed and exploited. We introduce a multi-phase partitioning, bringing heterogeneous control structures and heterogeneous software pipelining schemes to minimize control overheads and to maximize the efficiency of SWP. The proposal has been tested with 10 representative benchmarks extracted from baseband applications. In experiments cycle-counts, instructions, NOPs, LID/LIP access/miss/hit are thoroughly analyzed. Comparing to full FFTs with efficient SWP, our work reduces 20.5% - 87.5% cycle-counts, 11.2% - 86.5% instructions, 16.1% - 79.4% LID cache accesses and 19.5% - 87.1% LIP cache accesses. To the best of our knowledge, this is the first reported work about the generic software-pipelined PFFT on ILP architectures.


  • Systematic Optimization of Programmable QRD Implementation for Multiple Application Scenarios

    by Min Li, Javed Absar, Bruno Bougard, Liesbet van der Perre, Francky Catthoor

    Orthogonal-Triangular Decomposition (QRD) is one of the most fundamental signal processing primitives based on complex matrix operations [1]. It forms the core of many advanced multi-dimension and statistical signal processing algorithms that utilize orthogonalization, projection, and rank-revealing principles. Especially in the domain of wireless signal processing, many emerging algorithms in MIMO and OFDM systems have explicit or implicit connections to QRD [2]. This paper is about the systematic optimization of QRD implementation on programmable architectures. Based on the analysis of existing works, we introduce the following higher level components to the new optimization methodology: (1) Exploring high level algorithmic alternatives. (2) Categorizing different application scenarios. (3) Merging cascaded matrix operations. The systematic optimization brings significant improvements for programmable QRD implementations. Comparing to the widely accepted implementation in Numerical Receipts [3], our work achieves up to 79.76% cycle count reduction on TI TMS320C6713, a typical VLIW DSP. Moreover, our work achieves remarkable improvement on the memory subsystem, which is very critical for the power consumption and performance of modern DSP. Specifically, when QRD is used to solve least-square linear equations, our work reduces 99.55% LIP misses and 96.52% LID misses for 32×32 equations.

  • A Wavelet-FFT Based Efficient Sparse OFDMA Demodulator and Its Implementation on VLIW Architecture

    by Min Li, Bruno Bougard, David Novo, Liesbet van der Perre, Francky Catthoor

    In a practical OFDMA system, the terminal receiver often observes very low power in significant number of sub-carriers. The sparse energy distribution is because of the partial-load on sub-carriers, the beam-forming at basestation and the large amount of zero-sub-carriers obligated by communication standards. We present the algorithm and implementation of an energy-scalable OFDMA demodulator exploiting the afore-mentioned sparseness. The proposed scheme can dynamically reduce the computation-load and associated energy-consumption for OFDMA demodulator, which is one of the most energy-consuming parts of practical receivers. Our work applies Wavelet transform to perform decompositions based FFT, dropping low-energy bands and pruning useless dataflows. In addition, a new algorithm is proposed to significantly improve the accuracy of Wavelet-FFT with dropped bands. Our work has been implemented and tested on TI TMS320C6713, a typical VLIW baseband processor. In contrast to the well-studied partial FFT, our work is very friendly to the parallel execution on VLIW processor, enabling its application in contemporary programmable baseband implementations. In the case of 3GPP LTE receiver, 8.6% low-energy sub-carriers (in addition to the null sub-carriers) brings around 50% reduction of the cycle-count on TI TMS320C6713.

  • Efficient QRD for SRI-RLS Based Equalization on Programmable Architecture

    by Min Li, Bruno Bougard, Javed Absar, Francois Horlin, Liesbet Van Der Perre, Francky Catthoor

    Advanced adaptive filters have been shown to be very powerful for tracking time varying channels in various wireless communications system. However, the performance comes at the expense of highly resource-demanding implementations, especially in the context of programmable architecture based SDR. We present the optimizations for programmable implementation of QRD based SRI-RLS, which represents a large family of advanced adaptive filters. The key contribution of our work is to comprehensively and systematically remove the redundant operations in the QRD for SRI-RLS. Although most signal processing and scientific libraries implement householder reflection based QRD, we explore different alternatives and then choose given rotations based QRD to enable the aforementioned systematic redundancy removals. Our work significantly reduces the resource requirements (cycle count, energy consumption, etc.) of SRI-RLS implementation. Comparing to the widely accepted QRD implementation in numerical recipes, our work reduces 96.4% cycle-count on a typical baseband DSP (TI TMS320C6713), enabling efficient implementations. The paper shows that removing redundancy is very effective for modern statistical signal processing algorithms that largely rely on cascaded matrix operations.

  • Exploit Multiple-Domain Sparseness for HSDPA Chip Level Equalization in SDR: Algorithm and DSP Implementation

    by Min Li, Bruno Bougard, Francky Catthoor

    Chip level equalization has been proved as one of the key enabling technologies for HSDPA (high speed downlink packet access) receiver. Although many complicated algorithms (Kalman, etc.) have been reported to have great performance, their complexity and irregularity make it difficult to have efficient parallel software implementation. Targeting processor based SDR (software defined radio) platform, our goal is to design a practical HSDPA chip level equalizer having implementation cost as low as NLMS but offering considerable performance improvement over NLMS. Our proposal is based on the observations of multiple domain sparseness in cellular channel. The first observation is that the channel input response (CIR) has only a few significant taps. Although previous work exploits this for complexity reduction, we utilize it to improve the BER instead. The second observation is that the channel dynamics is not always significant, based on which we propose a feedback-control based technique to make the equalizer aware of the variation of channel dynamics. In addition, the equalizer becomes scalable in terms of quality-cost. By exploiting both of the aforementioned sparseness, the proposed HSDPA chip equalizer can significantly lower the BER error floor introduced by channel dynamics, so that more than 5 dB SNR gain can be achieved with the same implementation cost (by scalability) as NLMS. The design is demonstrated on TI TMS320C6711