Results 1 
6 of
6
Quantization Strategies For LowPower Communications
, 2001
"... Power reduction in digital communication systems can be achieved in many ways. Reduction of the wordlengths used to represent data and control variables in the digital circuits comprising a communication system is an effective strategy, as register power consumption increases with wordlength. Anothe ..."
Abstract

Cited by 6 (1 self)
 Add to MetaCart
Power reduction in digital communication systems can be achieved in many ways. Reduction of the wordlengths used to represent data and control variables in the digital circuits comprising a communication system is an effective strategy, as register power consumption increases with wordlength. Another strategy is the reduction of the required data transmission rate, and hence speed of the digital circuits, by efficient source encoding. In this dissertation, applications of both of these power reduction strategies are investigated. The LMS adaptive filter, for whichamyriad of applications exists in digital communication systems, is optimized for performance with a power consumption constraint. This optimization is achieved by an analysis of the effects of wordlength reduction on both performance  transient and steadystate  as well as power consumption. Analytical formulas for the residual steadystate mean square error (MSE) due to quantization versus wordlength of data and coefficient registers are used to determine the optimal allocation of bits to data versus coefficients under a power constraint. A condition on the wordlengths is derived under which the potentially hazardous transient "slowdown" phenomenon is avoided. The algorithm is then optimized for no slowdown and minimum MSE. Numerical studies are presented for the case of LMS channel equalization. Next, source encoding byvector quantization is studied for distributed hypothesis testing environments with simple binary hypotheses. It is shown that, in some cases, lowrate quantizers exist that cause no degradation in hypothesis testing performance. These cases are, however, uncommon. For the majority of cases, in which quantiza...
Incremental Communication for Multilayer Neural Networks: Error Analysis
, 1995
"... Artificial neural networks (ANNs) involve a large amount of internode communications. To reduce the communication cost as well as the time of learning process in ANNs, we earlier proposed an incremental internode communication method. In the incremental communication method, instead of communicati ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
Artificial neural networks (ANNs) involve a large amount of internode communications. To reduce the communication cost as well as the time of learning process in ANNs, we earlier proposed an incremental internode communication method. In the incremental communication method, instead of communicating the full magnitude of the output value of a node, only the increment or decrement to its previous value is sent on a communication link. In this paper, the effects of the limited precision incremental communication method on the convergence behavior and performance of multilayer neural networks are investigated. The nonlinear aspects of representing the incremental values with reduced (limited) precision for the commonly used error backpropagation training algorithm are analyzed. It is shown that the nonlinear effect of small perturbations in the input(s)/output of a node does not enforce instability. The analysis is supported by simulation studies of two problems. The simulation results ...
Echo Cancellation of Voiceband Data Signals Using RLS and StochasticGradient Algorithms
 IEEE Trans C‘ommzm. COM33
, 1985
"... AbstractThe convergence properties of adaptive least squares (LS) and stochastic gradient (SG) algorithms are studied in the context of echo cancellation of voiceband data signals. The algorithms considered are the SG transversal, SG lattice, LS transversal (fast Kalman), and LS lattice. It is show ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
AbstractThe convergence properties of adaptive least squares (LS) and stochastic gradient (SG) algorithms are studied in the context of echo cancellation of voiceband data signals. The algorithms considered are the SG transversal, SG lattice, LS transversal (fast Kalman), and LS lattice. It is shown that for the channel estimation problem considered here, LS algorithms converge in approximately 2N iterations where N is the order of the filter. In contrast, both SG algorithms display inferior convergence properties due to their reliance upon statistical averages. Simulations are presented to verify this result, and indicate that the fast Kalman algorithm NEAR ECHO frequently displays numerical instability which can be circumvented by HYBRID using the lattice structure. Finally, the equivalence between an LS algorithm and a fast converging modified SG algorithm which uses a maximum length input data sequence is shown. I.
et al
 IEEE Trans
, 1983
"... Abstract Adaptive channel equalization algorithms are commonly used in wireless communications receivers to counter intersymbol interference, multipath dispersion, and other time varying channel degradations. In this paper we obtain approximate expressions for the increase in mean square error of ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
Abstract Adaptive channel equalization algorithms are commonly used in wireless communications receivers to counter intersymbol interference, multipath dispersion, and other time varying channel degradations. In this paper we obtain approximate expressions for the increase in mean square error of the LMS adaptive algorithm when the total processing power is decreased by reducing the number of data and filter coefficient bits used by the algorithm. We also obtain expressions for the poweroptimal bitallocation factor which determines the proportion of the bits allocated to the data vs. allocated to the coefficients. Numerical studies are presented for an exponential memory IS1 channel and 4ary PSK signalling. These studies indicate that as few as 8 bits total are needed to equalize the channel and that most of these bits (6 out of 8) should be allocated to the filter coefficients. I.
Fast communication On the steadystate mean squared error of the fixedpoint LMS algorithm
, 2007
"... www.elsevier.com/locate/sigpro This communication studies the quantization effects on the steadystate performance of a fixedpoint implementation of the Least Mean Squares (LMS) adaptive algorithm. Based on experimental observations, we introduce a new intermediate mode of operation and develop a s ..."
Abstract
 Add to MetaCart
www.elsevier.com/locate/sigpro This communication studies the quantization effects on the steadystate performance of a fixedpoint implementation of the Least Mean Squares (LMS) adaptive algorithm. Based on experimental observations, we introduce a new intermediate mode of operation and develop a simplified theoretical approach to explain the behaviour caused by quantization effects in this mode. We also review the stall mode and provide a new expression that predicts the discontinuous behaviour of the steadystate mean squared error as a function of the input signal power. Combined with a previous analysis of quantization effects in stochastic gradient mode, this study provides analytical expressions for the steadystate mean squared error for the full range of stepsize values. We present experimental results that are in a good agreement with theoretical predictions to validate our model.
An LMS Based Algorithm for Reduced Finite Precision Effects
"... Abstract: This paper proposes a new Least Mean Square (LMS) based algorithm aimed for acoustic echo cancellation. The algorithm is an elaboration of an existing algorithm developed for the Konftel 200 conference phone. The purpose of the algorithm is to reduce the effects of quantization with only ..."
Abstract
 Add to MetaCart
Abstract: This paper proposes a new Least Mean Square (LMS) based algorithm aimed for acoustic echo cancellation. The algorithm is an elaboration of an existing algorithm developed for the Konftel 200 conference phone. The purpose of the algorithm is to reduce the effects of quantization with only a small increase in computational load as compared to the LMS algorithm. The algorithm makes use of the specific transfer characteristic of certain systems, e.g. handsfree phones, in combination with finite wordlength precision arithmetic. It is shown, both analytically and with simulations that the new algorithm outperforms the LMS algorithm when implemented in a two'scomplement arithmetic system identification scheme.