Neeraj Magotra

Lossless compression of waveform data for efficient storage and transmission

A two-stage technique for lossless waveform data compression is described. The first stage is a modified form of linear prediction with discrete coefficients, and the second stage is bilevel sequence coding. The linear predictor generates an error or residue sequence in a way such that exact reconstruction of the original data sequence can be accomplished with a simple algorithm. The residue sequence is essentially white Gaussian with seismic or other similar waveform data. Bilevel sequence coding, in which two sample sizes are chosen and the residue sequence is encoded into subsequences that alternate from one level to the other, further compresses the residue sequence. The algorithm is lossless, allowing exact, bit-for-bit recovery of the original data sequence. The performance of the algorithm at each stage is analyzed. Applications of the two-stage technique to typical seismic data indicates that an average number of compressed bits per sample close to the lower bound is achievable in practical situations. >

Single-station seismic event detection and location

A seismic event detection and source location (SEDSL) scheme which uses single-station (three-component) seismic data to analyze seismic event is presented. Each station monitors ground motion along three orthogonal directions-vertical, north, and east. To detect events, SEDSL combines the signals on the three components in a manner analogous to beam steering. Once an event is detected, SEDSL estimates the bearing of the source with respect to the receiver by estimating the polarization direction of the initial compressional phase. The range estimate is then obtained from the relative times of arrival of the different phases. >

Programmable digital hearing aid

A programmable customized universal digital listening system is provided with one or more digital signal processor chips which are implemented as one or more digital filters whose parameters are established by one or more erasable programmable read-only memories (EPROMs). The information included in the EPROMs directed to the parameters of the digital filters are determined based upon the users response to various audio signals provided from an audiologist. Based upon these responses, the EPROMs are programmed. Additionally, this listening system is provided with an additional digital filter which changes its responses based upon the frequency of any background noise.

Lossless predictive coding

This paper describes the first stage of a two stage lossless data compression algorithm. The first stage consists of a lossless adaptive predictor. The term lossless implies that the original data can be recovered exactly. The second stage employs arithmetic coding. Results are presented for a seismic data base.

A comparison of two parameter estimation schemes

This letter compares two parameter estimation schemes using the variance of the estimate as the criterion for comparison.

Lossless seismic data compression using adaptive linear prediction

This paper presents a comparison of adaptive linear predictors as applied to the area of lossless compression of seismic waveform data. Three methods are explored: the normalize least-mean square (NLMS) algorithm, the gradient adaptive lattice (GAL) algorithm, and the recursive least squares lattice (RLSL) algorithm. When compared to standard linear prediction techniques, all three of these methods require little overhead, are more computationally efficient, and can be implemented using floating point techniques. With respect to a standard seismic database, the RLSL filter outperforms the other two methods in nearly all cases tested.

Effects of multiple-pass filtering in lossless predictive compression of waveform data

Presents the effects of the predictive filtering of waveform data in multiple passes as the first stage of a two-stage lossless compression algorithm. Predictive compression has a proven track record when applied to high dynamic range waveform data, wherein the waveform data are input to a linear predictor or perhaps an adaptive predictor for decorrelation, and the resultant residue is then subjected to an entropy coder to (ideally) represent the signal with a minimum number of bits. This compression is commonly applied with no loss of information. In this work, an adaptive filter is used for prediction, but instead of a single run through the predictor, the residue is continually passed back through the predictor in an attempt to further decorrelate the residue. Multiple passes of a gradient adaptive lattice filter has given the best decorrelation, yielding improved compression ratios. We run the compression technique on a seismic database, then provide some comparative lossless compression results using several coding schemes and show that using multiple-pass predictive filtering can improve the compression rates attainable.

Low power real-time programmable DSP development platform for digital hearing aids

This paper presents a new low power binaural wearable digital hearing aid platform based on the Texas Instruments TMS320C5000 fixed point digital signal processor. This platform is a real-time system capable of processing two input speech channels at a 32 kHz sampling rate for each channel and driving a stereo headphone output. It provides for frequency shaping, suppression, multiband amplitude compression, frequency dependent interaural time delay algorithms. Since the platform is a programmable solution capable of running at 1.8 V for MIPS intensive research and 1 V for actual hearing aid implementation, this platform will enable further research into improving the quality of life for the hearing impaired.

Interactive DSP course development/teaching environment

The authors have developed an interactive environment for the creation and maintenance of dynamic, active multimedia-based teaching mechanisms. The environment is designed to be user-friendly and to facilitate the creation of educational material. This tool has already been used to create courses in multidimensional signal processing (MDSP) by researchers working together while geographically separated.

Lossless compression of electroencephalographic (EEG) data

The lossless compression of electroencephalographic (EEG) data is of great interest to the biomedical research community. In this paper, a two-stage technique of lossless compression involving decorrelating the sample points of the EEG signal and then entropy coding the resulting signal is examined. Two alternatives are presented for performing the first task. Specifically, the first stage consists either of a fixed coefficient filter or a recursive least squares lattice filter. The second stage employs arithmetic coding to perform the task of entropy coding the data. In the decompression stage, exact inverse filters are applied to achieve lossless compression. Simulations demonstrate the feasibility of this method for lossless EEG data compression.