Signal Processing Application

To implement any  Signal Processing operation on one dimensional Signal.

It was a group experiment .We had to find out an application of DSP for 1-dimensional signal.Our group members are Apurva Harane ,Anuradha Kadam ,Seema Bhoir and me.We selected ECG ie Electrocardiography as the application of DSP and the application of ECG which I have selected is ECG  derived sleep monitoring .

  APPLICATION: Method and apparatus for ECG derived sleep monitoring of a user 

 Publication number : US20050148895 A1
Application number : US 11/020,953
Publication date : Jul 7, 2005
Filing date : Dec 22, 2004
 Inventors : Dale Misczynski, Vladislav Bukhman

Summary:
The invention relates to a method and device for simultaneous monitoring of cardiac activity, respiratory rate, and the evaluation of sleep states of a user. The invention comprises a seamless on-line evaluation of cardiac activity, heart rate variability, and ECG derived respiratory rate and using these data as an input for evaluation and quantitative assessment of sleep disorders. The hardware comprises of a flexible cardiac belt with 3 electrodes, which is easy to use without involvement of trained professionals.
To facilitate the diagnosis of sleep disorders, patients are monitored using polygraph recording of electroencephalograms (EEG), electrocardiogram (ECG), electro-oculogram (EOG) and other data.
The fundamental aspect of the invention is a simultaneous monitoring of ECG and respiratory parameters, evaluation of autonomic nervous system (ANS) activity, and determining of sleep states incorporated in a wearable unobtrusive device which can be used for in-home, ambulatory and in-hospital sleep monitoring without supervision or assistance of trained personal.

https://www.google.com/patents/US20050148895

http://www.hindawi.com/journals/isrn/2012/768794/

Basic Operations using DSP Processor

This was the first hardware based demo of signal processing. The session was on TMS320F28375 DSP board. Emulation was observed of real time audio input in class demonstration.
 We learned to perform basic arithmetic functions-

1. Basic Arithmetic operations - ADD,SUB,MUL,DIV.
2. Bitwise operations - AND, NOT
3. Shifting operation - SHIFT LEFT/RIGHT AND ROTATE LEFT/RIGHT

It was concluded that a large number of mathematical operations can be performed repeatedly on this processor as it is a specialised microprocessor.

Design FIR Filter using Frequency Sampling Method

The objective of this experiment is to design the digital filter using frequency sampling method.

The basic concept of frequency sampling is sampling the desired frequency response H_d (ω) at          ω=(2πk/N). Here, the desired frequency response H_d (ω) is ideal filter magnitude which is multiplied  by linear phase. The phase response obtained by frequency sampling is same for both LPF and HPF. Ripples are obtained in the stop band of decreasing amplitude.

https://drive.google.com/drive/folders/0ByMxeH1FMa8uNWtsOV8yUkcxYlk

Design of F I R Filter

The objective is to design the digital Linear phase FIR filter using windowing technique and study the spectrum of the filter.

"Linear Phase" refers to the condition where the phase response of the filter is a linear (straight-line) function of frequency . This results in the delay through the filter being the same at all frequencies. Therefore, the filter does not cause "phase distortion" or "delay distortion".

https://drive.google.com/drive/folders/0ByMxeH1FMa8uWFRvdDFVd1FoSEk

Design of Chebyshev IIR Filter

The aim of the experiment is to design Analog and Digital Chebyshev Filter design.

 Input Specifications given are :

For LPF / HPF filter Design :

(1) Pass band Attenuation (Ap) ,  (2) Stop band Attenuation (As ),

(3) Pass band Frequency (Fp) in Hz  ,(4) Stop band Frequency (Fs) in Hz,

(5) Sampling Frequency in Hz.

The digital Chebyshev filter was designed using the same procedure as the Butterworth Filter using Scilab.The characteristics show the ripple behavior in pass band and monotonic in stop band.

 The number of peak and valleys in the pass band determine the order of the filter.

The basic difference between the Butterworth and chebyshev filter is that the poles of Butterworth filters lie on a circle and that of Chebyshev filter on the ellipse. It is because of their different characterics.

Chebyshev filters are analog or digital filters having a steeper roll-off and more pass band ripple (type I) or stop band ripple (type II) than Butter worth filters.

https://drive.google.com/drive/folders/0ByMxeH1FMa8uQm10b0dVeUVnNTQ

Design of Butterworth Filter

The aim of the experiment is to design Analog and Digital Butterworth Filter.

The objective is to design a digital filter from Analog filter and study the aliasing effect due to sampling in Impulse Invariant Method and the frequency warping effect in BLT Method.

 Input Specifications :

(1) Pass band Attenuation (Ap< 3 dB)  (2) Stop band Attenuation (As> 40 dB )

(3) Pass band Frequency (Fp) in Hz  (4) Stop band Frequency (Fs) in Hz

(5) Sampling Frequency (F) in Hz

In this experiment we have studied that,

1.  response in butterworth filter is monotonic. 

2. As the order increases the response becomes more sharpen in both high pass filter and low pas filter.

3. In this designing we initially find order through pass band and stop band frequency.and find normalized and denormalised transfer function and the find actual transfer function in z transform.

https://drive.google.com/drive/folders/0ByMxeH1FMa8uYW5Ecnl4QWJiZkk

Filtering of long Data Sequence

The aim of the experiment is to perform filtering of Long Data Sequence using Overlap Add Method and Overlap Save Method. 

The objective is to implement filtering of Long Input Sequence using Overlap Add / 

Overlap Save Algorithm.

The input specifications given are:

 1) Length of long data sequence and Signal values. 

2) Length of impulse response M and Signal values.

We concluded that Overlap Add Method and Overlap Save Method are much suitable for filtering Long Data Sequence. 

The link for above is:

https://drive.google.com/drive/folders/0ByMxeH1FMa8uakhuX3hoX2Y3LWs

Fast Fourier Transform

The aim of the experiment is to perform Fast Fourier Transform.

The objective is to develop a program to perform FFT of N point signal.

The input specifications were  Length of Signal N and Signal values.

Fast Fourier Transform reduces the number of computations required as compaired to by DFT.

We take any four-point sequence x[n] and find X[k] using forward FFT. 

In FFT, the number of computation done is in the power of 2 to accelerate computation.

https://drive.google.com/drive/folders/0ByMxeH1FMa8uRUpfRjdlcmZJZ1E

Discrete Fourier Transform

Learning experience :

The objective of the experiment is to develop a function to perform Discrete Fourier Transform of N point Signal and conclude the effect of zero padding on magnitude spectrum.

Discrete Fourier Transform is used to transform the signal from time domain to frequency domain.
We observed that as the value of N increases,the spacing between the values on the magnitude spectrum reduces and hence ,the resolution increases and the approximation error reduces.

We performed IDFT also to get the original signal back.

  

https://drive.google.com/drive/folders/0ByMxeH1FMa8uOFlkcm9RS0RhcDQ

Linear convolution, Circular convolution, Linear convolution using circular convolution and Correlation

Learning experience :

  The objective of the experiment was to Develop a function to find Linear Convolution and Circular Convolution, Calculate Linear Convolution, Circular Convolution, Linear Convolution using Circular Convolution and verify the results using mathematical formulation and Conclude on aliasing effect in Circular convolution.

   The Input Specifications given were :  Length of first Signal L and Signal values, Length of second Signal M and Signal values.

  We performed the experiment in C language and concluded that:

1. If both the input signals are causal then output is also causal.

2.In Linear convolution, the length of output signal is L+M-1, where L is length of x[n] and M is length of h[n].

3. Circular convolution gives aliased output that is the output of linear convolution gets overlapped and gives aliased output.

4.Correlation is used to find the degree of similarities between the two signals. For example as used in RADAR. 

The link for above is :https://drive.google.com/drive/folders/0ByMxeH1FMa8uOWFzbWxSUlB4elE