Conference Paper

High-speed FPGA implementation of an improved LMS algorithm

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

The FPGA implementation of a new parallel processing method is studied by introducing the parallel processing method into the delayed least mean square (DLMS) algorithm. The parallel delayed least mean square (PDLMS) algorithm has the faster data throughput and higher convergence rate than the DLMS algorithm. In this paper, the hardware implementation of PDLMS is realized by hardware description language, while the simulation structure is presented. The results show that the PDLMS algorithm has certain superiority according to DLMS.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Various implementation of the adaptive filter using LMS algorithm is done in FPGA [4,6,[9][10][11]. LMS algorithm is implemented in VLSI using various architectures [6]. ...
Article
Full-text available
Adaptive noise cancellation is an extensively researched area of signal processing. Many algorithms had been studied such as least mean square algorithm (LMS), recursive least square algorithm, and normalized LMS algorithm. The statistical characteristics of noise are fast in nature and the algorithms for noise cancellation should converge fast. Since LMS algorithm has slow convergence; in this paper, a variable leaky LMS (VLLMS) algorithm is explored. VLLMS is implemented using the concept of hardware-software cosimulation using Xilinx System Generator. The design is implemented on Virtex-6 ML605 field programmable gate array board. The implemented design is tested for sinusoidal signal added with an additivewhite Gaussian noise. The design summary and the utilization summary are presented.
... An efficient method to determine the delays in the DLMS filter [5].Different methods will allow a designer to explore trade-off at a highlevel in the design flow. The hardware implementation of PDLMS is realized by hardware description language, while the simulation structure is presented [6]. Analyzed the convergence performance and data throughput of PDLMS algorithm which has certain superiority according to DLMS. a stochastic analysis of the delayed leastmean-square (DLMS) adaptive algorithm with leakage [7]. ...
Conference Paper
Full-text available
In this paper discussed about improving the architectural efficiency in Delayed Mean Square Filter. Based on the partial product generators and binary adder tree structures of filter blocks to achieving the low area, delay, power products using error computational block and weight update block in Delayed Mean Square Filter. The new parallel processing method is studied by introducing the parallel processing method into the delayed least mean square (DLMS) algorithm. Accuracy of our model is analyzed by simulations.
Conference Paper
Full-text available
RESUMEN: En este trabajo se describe la implementación de un algoritmo de filtrado adaptativo basado en características estadísticas de la señal, utilizando como función de coste el criterio del valor esperado de la señal, no así su valor absoluto. En el trabajo se realiza además un análisis de la implementación descrita en cuanto a consumo de recursos hardware, velocidad de convergencia y su comparación con el criterio LMS y LMK, para distintos niveles de ruido. Tomando como eje fundamental del trabajo, la flexibilidad de los dispositivos FPGAS y las ventajas de la aceleración de algoritmos por hardware en determinadas aplicaciones. El filtro desarrollado fue verificado experimentalmente en una aplicación de reducción de ruido en señales, con resultados positivos Palabras Clave: Varianza, Cancelación de Ruido, Filtrado Adaptativo. ABSTRACT: This work describes the implementation of an adaptive filtering algorithms based on statistical characteristics of signals , using as the cost function, the signal expected value criteria instead of its absolute value. Furthermore, a comparison of the implementation in terms of hardware resource consumption, and convergence speed is made. Besides of its compare with the results of LMS and LMK algorithms for different noise levels. Taking as central axis of the work, the flexibility of FPGAS devices and the advantages of algorithms acceleration for hardware in specific applications. The developed filter was verified in a noise reduction application, with highly satisfactory results.
Article
A variation of the least means squares (LMS) algorithm, called the delayed LMS (DLMS) algorithm is ideally suited for highly pipelined, adaptive digital filter implementations. In this paper, we present an efficient method to determine the delays in the DLMS filter. Furthermore, in order to achieve fully pipelined circuit architectures for FPGA implementation, we transfer these delays using retiming. The method has been used to derive a series of retimed delayed LMS (RDLMS) architectures, which allow a 66.7% reduction in delays and 5 times faster convergence time thereby giving superior performance in terms of throughput rate when compared to previous work. Three circuit architectures and three hardware shared versions are presented which have been implemented using the Virtex-II FPGA technology resulting in a throughput rate of 182 Msample/s.
Article
A novel parallel delayed least-mean-square (PDLMS) algorithm is proposed by introducing the parallel processing method into delayed LMS (DLMS) algorithm. Compared with DLMS, the algorithm presented has the property of smaller delay, higher throughput rate and faster convergence speed, while it also exhibits some de-correlation effect for the correlated input sequence. These properties make it more suitable for the cases of high order filter with high convergence speed. At the same time, it can be mapped onto the high-speed and/or high-pipelined hardware structure directly.
Conference Paper
This paper proposed a fixed-point implementation method of LMS (least mean square) and N-LMS (normalized-LMS) processor. In N-LMS, this paper proposes an efficient method using simple bit-shift operation instead of division. The convergence performance in LMS, N-LMS and RLS (recursive least square) adaptive array antenna is compared by implementation with single large scale FPGA (field programmable gate array) on the same developed hardware platform. It was evaluated by using the actual processing time considering the operation clock speed instead of the number of weight updates. The fixed-point operation with optimized word length and bit-shift operation instead of division are expected to provide faster actual FPGA processing time for LMS families compared with RLS in some specific cases
Conference Paper
The high-throughput delayed LMS (DLMS) adaptive algorithm suffers from a slower convergence rate compared to the LMS algorithm. Different versions of the DLMS adaptive algorithm using a conversion scheme have been proposed to improve the convergence rate. This improved convergence was achieved at the expense of an increased computational complexity and a lower throughput rate than the original DLMS algorithm. We propose a new modified DLMS adaptive algorithm that, compared to the existing conversion-based DLMS algorithm, provides a higher throughput rate for a similar convergence rate. Alternatively, the proposed algorithm provides a faster convergence for the same throughput rate compared to the conversion-based DLMS algorithm. In both the cases, the computational complexity of the proposed algorithm is smaller than that of the conversion-based DLMS algorithm. The proposed algorithm uses the error signal from each stage of the adaptive FIR filter independently to update the value of the corresponding coefficient. Simulations illustrate the convergence performance of the new algorithm. The performance of its architecture is evaluated in terms of computational complexity, throughput, and latency. The proposed algorithm provides a better throughput rate and a computational complexity lower than that of the conversion-based DLMS algorithm.
Article
High-speed field-programmable gate array (FPGA) implementations of an adaptive least mean square (LMS) filter with application in an electronic support measures (ESM) digital receiver, are presented. They employ "fine-grained" pipelining, i.e., pipelining within the processor and result in an increased output latency when used in the LMS recursive system. Therefore, the major challenge is to maintain a low latency output whilst increasing the pipeline stage in the filter for higher speeds. Using the delayed LMS (DLMS) algorithm, fine-grained pipelined FPGA implementations using both the direct form (DF) and the transposed form (TF) are considered and compared. It is shown that the direct form LMS filter utilizes the FPGA resources more efficiently thereby allowing a 120 MHz sampling rate.
Article
The problem of implementing a high sampling rate transversal form adaptive filter is investigated. A highly pipelined systolic-type alternative to the conventional LMS adaptive filter is presented. The proposed system consists of a linear array of identical processing modules specifically suited to the computational requirements of the delayed LMS algorithm. The resulting adaptive filter structure can accommodate very high sampling rates, which are independent of the filter order. The performance of the system is analyzed in terms of computational speedup and maximum sampling rate, and the effect of adaptation delay on algorithm convergence is addressed
Verilog HDL: A Guide to Digital Design and Synthesis Second Edition
  • Xia Yuwen
  • Hu Yanxiang
  • Diao Lansong
PDLMS algorithm realization based on multi-DSP system
  • Jiang Sha
  • Tian Jiansheng