Unit II: Basic Instrumentation

Types of data and Role of DSP if collected data

contain Noise

Dr. Krishna K. Warhade

Ph.D. IIT Bombay
Dean Research and Head PhD Research Centers,
MITWPU,Pune
1/22/2022 1
 Classification of data

◼ Secondary data
◼ Primary data

1/22/2022 2
Secondary Data – Examples of Sources
◼ County health departments
◼ Vital Statistics – birth, death certificates
◼ Hospital, clinic, school nurse records
◼ Private and foundation databases
◼ City and county governments
◼ Surveillance data from state government
programs
◼ Federal agency statistics - Census, NIH, etc.
◼ Fingerprint , face, iris data
Secondary Data – Limitations
◼ When was it collected? For how long?
◼ May be out of date for what you want to analyze.
◼ May not have been collected long enough for
detecting trends.
◼ Is the data set complete?
◼ There may be missing information on some
observations
◼ Unless such missing information is caught and
corrected for, analysis will be biased.
Secondary Data – Limitations
◼ Are there confounding problems?
◼ Sample selection bias?
◼ Source choice bias?
◼ In time series, did some observations drop out over
time?
Secondary Data – Advantage

◼ It will save you money.

◼ Even if you have to pay for access, often it is

cheaper in terms of money than collecting your

own data.
◼ It will save you time.
◼ Primary data collection is very time consuming.

◼ It may be very accurate.

◼ When especially a government agency has

collected the data, incredible amounts of time and

money went into it. It’s probably highly accurate.
Primary Data - Examples
◼ Surveys

◼ Focus groups

◼ Questionnaires

◼ Personal interviews

◼ Experiments and observational study

 Primary Data - Limitations

◼ Do you have the time and money for:

◼ Designing your collection instrument?

◼ Selecting your population or sample?
◼ Pretesting/piloting the instrument to work out sources
of bias?
◼ Administration of the instrument?
◼ Entry/collation of data?
 Primary Data - Limitations
◼ Uniqueness
◼ May not be able to compare to other populations

◼ Researcher error
◼ Sample bias

◼ Other confounding factors

Data collection choice
◼ If that data exist in secondary form, then use
them to the extent you can, keeping in mind
limitations.

◼ But if it does not, and you are able to fund

primary collection, then it is the method of
choice.
Methods of data Collection
Collection of Primary data

◼ Observation Method

◼ Interview Method

◼ Through Questionnaires

◼ Through Schedules
 Observation Method
◼ The information is sought by way of investigators own
direct observation without asking from the respondents.
◼ Subjective bias is eliminated if observation is done
accurately .
◼ Information obtained is current, not affected by past.
◼ Independent of respondents willingness.
◼ Limitation: Expensive, limited information

◼ Participant Observation, Non-Participant observation

◼ Controlled observation, Uncontrolled observation
 Interview Method:
◼ Personal interview.
Direct personal investigation or indirect oral
investigation.
(structured interview, Nonstructural interview)

◼ Telephonic Interview
Collecting information through telephone
 Interview Method:
Personal interview.
Merits
◼ Greater depth information obtained

◼ Greater flexibility, samples can be controlled

◼ Can collect supplementary information

Demerits
◼ Expensive, bias interview

◼ Time consuming
 Telephonic Method:
Merits
◼ More flexible, Higher rate of response

◼ Faster, cheaper

◼ Replies can be recorded

Demerits
◼ Little time is given to respondents

◼ Restricted to telephone facility at other end

◼ Questions have to be short and to the point

 Collection of data through Questionnaires:
Popular and used by private & public organization,
government
Merits
◼ Low cost

◼ Free from bias, Adequate time for respondents.

◼ Large samples can be collected.

Demerits
◼ Low rate of return of form.

◼ Only Educated people response

◼ Possibility of Ambiguous replies or omission

◼ Slowest of all
 Collection of data through Schedules:
Performa containing set of questions are being filled by
Persons who are specially appointed (Population census)

◼ Persons should be trained.

◼ Person should be honest and sincere.
◼ Lead to fairly reliable results.
◼ Very expensive.
 Secondary data
◼ Publication of central, state government.
◼ International bodies.
◼ Technical journals, books, magazines, newspaper.
◼ Reports of business, industry, banks, stock exchange.

Before using secondary date, ensure

◼ Reliability of data
◼ Suitability of data
◼ Adequacy of data
 Selection of appropriate method for data
collection

◼ Nature scope and object of inquiry.

◼ Availability of funds.

◼ Time factor

◼ Precision required
 SAMPLING

◼ A sample is “a smaller (but hopefully representative)

collection of units from a population used to determine
truths about that population”
◼ Why sample?
◼ Resources (time, money) and workload

◼ Gives results with known accuracy that can be

calculated mathematically
◼ The sampling frame is the list from which the potential
respondents are drawn
◼ Registrar’s office

◼ Class rosters

◼ Must assess sampling frame errors

20
SAMPLING……

◼ What is your population of interest?

◼ To whom do you want to generalize your results?

◼ All doctors

◼ School children

◼ Indians

◼ Women aged 15-45 years

◼ Other

◼ Can you sample the entire population?

21
 SAMPLING…….

◼ Three factors that influence sample

representativeness
◼ Sampling procedure
◼ Sample size
◼ Participation (response)

◼ When might you sample the entire population?

◼ When your population is very small
◼ When you have extensive resources
◼ When you don’t expect a very high response

22
SAMPLING…….

STUDY POPULATION

SAMPLE

TARGET POPULATION

23
 Types of Samples

◼ Probability (Random) Samples

◼ Simple random sample
◼ Systematic random sample

◼ Stratified random sample

◼ Multistage sample

◼ Multiphase sample

◼ Cluster sample

◼ Non-Probability Samples
◼ Convenience sample

◼ Purposive sample

◼ Quota
24
 Process

◼ The sampling process comprises several stages:

◼ Defining the population of concern

◼ Specifying a sampling frame, a set of items or events

possible to measure
◼ Specifying a sampling method for selecting items or

events from the frame

◼ Determining the sample size

◼ Implementing the sampling plan

◼ Sampling and data collecting

◼ Reviewing the sampling process

25
 Population definition

◼ A population can be defined as including all people or

items with the characteristic one wishes to understand.

◼ Because there is very rarely enough time or money to

gather information from everyone or everything in a
population, the goal becomes finding a representative
sample (or subset) of that population.

26
 Population definition…….

◼ Note also that the population from which the sample is

drawn may not be the same as the population about which
we actually want information. Often there is large but not
complete overlap between these two groups due to frame
issues etc .

◼ Sometimes they may be entirely separate - for instance, we

might study rats in order to get a better understanding of
human health, or we might study records from people born
in 2008 in order to make predictions about people born in
2009.
27
 SAMPLING FRAME

◼ In the most straightforward case, such as the sentencing of a

batch of material from production (acceptance sampling by lots), it
is possible to identify and measure every single item in the
population and to include any one of them in our sample.

◼ However, in the more general case this is not possible. There is

no way to identify all rats in the set of all rats. Where voting is not
compulsory, there is no way to identify which people will actually
vote at a forthcoming election (in advance of the election)

◼ As a remedy, we seek a sampling frame which has the

property that we can identify every single element and
include any in our sample .
◼ The sampling frame must be representative of the population
28
 PROBABILITY SAMPLING

◼ A probability sampling scheme is one in which every

unit in the population has a chance (greater than zero)
of being selected in the sample, and this probability can
be accurately determined.

◼ When every element in the population does have the

same probability of selection, this is known as an 'equal
probability of selection' (EPS) design. Such designs are
also referred to as 'self-weighting' because all sampled
units are given the same weight.
29
 PROBABILITY SAMPLING…….

Probability sampling includes:

◼ Simple Random Sampling,
◼ Systematic Sampling,
◼ Stratified Random Sampling,
◼ Cluster Sampling
◼ Multistage Sampling.
◼ Multiphase sampling
30
 NON PROBABILITY SAMPLING

◼ Any sampling method where some elements of population

have no chance of selection (these are sometimes referred
to as 'out of coverage'/'undercovered'), or where the
probability of selection can't be accurately determined.

◼ It involves the selection of elements based on assumptions

regarding the population of interest, which forms the criteria
for selection. Hence, because the selection of elements is
nonrandom, nonprobability sampling not allows the
estimation of sampling errors..

31
 NONPROBABILITY SAMPLING…….
• Nonprobability Sampling includes: Accidental
Sampling, Quota Sampling and Purposive Sampling.

• In addition, nonresponse effects may turn any

probability design into a nonprobability design if the
characteristics of nonresponse are not well
understood, since nonresponse effectively modifies
each element's probability of being sampled.

32
 SIMPLE RANDOM SAMPLING

• Applicable when population is small, homogeneous & readily

available
• All subsets of the frame are given an equal probability. Each
element of the frame thus has an equal probability of
selection.
• It provides for greatest number of possible samples. This is
done by assigning a number to each unit in the sampling
frame.
• A table of random number or lottery system is used to
determine which units are to be selected.

33
 SIMPLE RANDOM SAMPLING……..

◼ Estimates are easy to calculate.

◼ Simple random sampling is always an EPS (Equal
probability of selection) design, but not all EPS designs
are simple random sampling.

◼ Disadvantages
◼ If sampling frame large, this method impracticable.
◼ Minority subgroups of interest in population may not be
present in sample in sufficient numbers for study.

34
 SYSTEMATIC SAMPLING

◼ Relies on arranging the target population according to some ordering

scheme and then selecting elements at regular intervals through that
ordered list.

◼ Systematic sampling involves a random start and then proceeds with

the selection of every kth element from then onwards. In this case,
k=(population size/sample size).

◼ It is important that the starting point is not automatically the first in

the list, but is instead randomly chosen from within the first to the
kth element in the list.

◼ A simple example would be to select every 10th name from the

telephone directory (an 'every 10th' sample, also referred to as
'sampling with a skip of 10').
35
 SYSTEMATIC SAMPLING……

systematic sampling is an EPS method, because all elements have the

same probability of selection.

36
 SYSTEMATIC SAMPLING……

ADVANTAGES:
◼ Sample easy to select
◼ Suitable sampling frame can be identified easily
◼ Sample evenly spread over entire reference population

DISADVANTAGES:
◼ Sample may be biased if hidden periodicity in population
coincides with that of selection.
◼ Difficult to assess precision of estimate from one survey.

37
 STRATIFIED SAMPLING

Where population embraces a number of distinct categories, the frame

can be organized into separate "strata." Each stratum is then sampled as
an independent sub-population, out of which individual elements can be
randomly selected.
◼ Every unit in a stratum has same chance of being selected.

◼ Using same sampling fraction for all strata ensures proportionate

representation in the sample.

◼ Adequate representation of minority subgroups of interest can be ensured

by stratification & varying sampling fraction between strata as required.

38
STRATIFIED SAMPLING…….

Draw a sample from each stratum

39
 CLUSTER SAMPLING

◼ Cluster sampling is an example of 'two-stage sampling' .

◼ First stage a sample of areas is chosen;
◼ Second stage a sample of respondents within those areas is
selected.
◼ Population divided into clusters of homogeneous units,
usually based on geographical contiguity.
◼ Sampling units are groups rather than individuals.
◼ A sample of such clusters is then selected.
◼ All units from the selected clusters are studied.

40
 CLUSTER SAMPLING…….
Advantages :
◼ Cuts down on the cost of preparing a sampling
frame.
◼ This can reduce travel and other administrative costs.

Disadvantages:
sampling error is higher for a simple random sample
of same size.
◼ Often used to evaluate vaccination coverage in EPI
(Expanded Program on Immunization)

41
 MULTISTAGE SAMPLING

◼ Complex form of cluster sampling in which two or more

levels of units are embedded one in the other.

◼ First stage, random number of districts chosen in all

states.

◼ Followed by random number of talukas, villages.

◼ Then third stage units will be houses.

◼ All ultimate units (houses, for instance) selected at last

step are surveyed.
42
 MULTISTAGE SAMPLING……..
◼ This technique, is essentially the process of taking random samples of
preceding random samples.

◼ Not as effective as true random sampling, but probably solves more of

the problems inherent to random sampling.

◼ An effective strategy because it banks on multiple randomizations. As

such, extremely useful.

◼ Multistage sampling used frequently when a complete list of all members

of the population not exists and is inappropriate.

◼ Moreover, by avoiding the use of all sample units in all selected clusters,
multistage sampling avoids the large, and perhaps unnecessary, costs
associated with traditional cluster sampling. 43
 Non probability sampling:
QUOTA SAMPLING

◼ The population is first segmented into mutually exclusive sub-groups,

just as in stratified sampling.
◼ Then judgment used to select subjects or units from each segment
based on a specified proportion.
◼ For example, an interviewer may be told to sample 200 females and 300
males between the age of 45 and 60.
◼ It is this second step which makes the technique one of non-probability
sampling.
◼ In quota sampling the selection of the sample is non-random.
◼ For example interviewers might be tempted to interview those who look
most helpful. The problem is that these samples may be biased because
not everyone gets a chance of selection. This random element is its
greatest weakness and quota versus probability has been a matter of
controversy for many years
44
 Non probability sampling:
Purposive sampling
◼ The researcher chooses the sample based on
who they think would be appropriate for the
study. This is used primarily when there is a
limited number of people that have expertise
in the area being researched

45
 Inferential Statistics

◼ Estimation
◼ e.g., Estimate the population
mean weight using the sample
mean weight
◼ Hypothesis testing
◼ e.g., Test the claim that the
population mean weight is 70
kg
Inference is the process of drawing conclusions or making
decisions about a population based on sample results
Role of DSP if data contain noise
Sources of noise

◼ Operational reliability of system mostly depends on the

environmental factors called as Noise
◼ Noise can be Random in nature, repetitive, continuous,
independent burst
◼ To eliminate or reduce noise it is important to understand the
sources of noise.
◼ Sources of noise include internal sources, external sources and
locally produced noise
Sources of noise

Internal Noise sources: This type of noise is generated by

components of electronics system. Following are some examples of
noise generated by internal noise sources
◼ White Noise: This type of noise affects equally all the
frequencies. It has gaussian amplitude distribution
◼ Shot noise (Poisson noise): It is a type of electronic noise which
can be modelled by a poisson process. Shot noise originates
from the discrete nature of electric charge.
◼ Thermal noise: This is the electronic noise generated by the
thermal agitation of the charge carriers inside an electrical
conductor at equilibrium, which happens regardless of any
applied voltage.
◼ Flicker noise: This is a type of noise with a 1/f power spectral
density. Noise increases with decrease in frequency.
Sources of noise

▪ Barkhausen noise: If the system contains some magnetic

sensors then due to finite size of domain of ferromagnetic
materials the noise get introduced.
▪ Contact noise: When there is breakdown of contacts between
adjacent particles in current path the noise introduced is called
as contact noise
Sources of noise:
External noise sources: This type of noise is caused due to
external factors or due to other electrical systems
▪ Switching current and voltages: When high current load switches turns
on and off they cause transients in power supply lines because of their
inductance and capacitance.
▪ Power line interference: When a high current is flowing through the
line parallel to signal line it provides noise even if it is few distance
away.
▪ Sparking and Radiation: Electromagnetic waves as a noise is produced
whenever there is spark. Unshielded lines and open conductors are
also causes noise.
▪ Environmental and atmospheric noise: Lightening discharge in
thunderstorm causes spurious radio waves in the form of impulses that
affects radio broadcasting
▪ Electrostatic discharge: This serious problem observed in areas
Where there is dryer air, air condition and continental climates
 Sources of noise
Local noise sources :

This type of noise is associated with circuit interconnections, broad

interfacing, transmission cables and power supplies. This can be
reduced by providing attention while designing the circuits.

1/22/2022 52
 DSP application in noise filtering
◼ Geophysical data signal processing
◼ Structural health monitoring in civil engineering
◼ Medical instrumentation
◼ Seismology
◼ Genomics
◼ Digital filtering
◼ DSP in ECG
◼ Signal processing application in Automotive
◼ Satellite image processing
◼ Biomedical signal and image processing
◼ Defense

1/22/2022 53
ECG analysis: Why filtering is required?
◼ Most biomedical signals are weak signals.
◼ Any signal other than that of interest must be considered as
interference, artifact, or noise.
◼ Artifacts degrade the performance of signal processing
algorithms.
◼ Due to weak nature of biomedical signals at their source, high
amplification factor is required.
◼ But Electronic noise in instrumentation amplifier also gets
amplified with desired signal (Random noise).
◼ EM waves broadcast by radio, television stations, mobile towers,
computer monitors, system used in Labs are picked up by
cables, devices and connectors.
◼ 50 Hz power supply waveform can also get mixed with signals
and corrupt signal of interest (Structured noise).
Prerequisites for filtering
◼ Background in signals and systems, Convolution and correlation.
◼ Probability and random variables.
◼ Stationary and Nonstationary signals (EMG, EEG, VMG, PCG)
◼ Laplace transform, Fourier Transform, Z-Transform.
◼ DFT, FFT.
◼ Design of filters- time domain, frequency domain- IIR/FIR.
◼ Time domain filter like moving averaging filters and derivative
filters required Z.T. knowledge
◼ Butterworth and chebyshev LP/HP/BP/BR filter design required.
◼ Knowledge of optimal filtering such as wiener filter required.
◼ Adaptive filters knowledge for removal of artifacts and adaptive
cancellation of the maternal ECG to obtained the Fetal ECG.
◼ Adaptive cancellation of Muscle-contraction interference in
Knee-joint vibration signals
 Physiological Interference

➢ EMG related to coughing, breathing affects ECG

➢ EGG interfering with precordial ECG

➢ Maternal ECG added to the fetal ECG of Interest

➢ ECG interfering with the EEG

➢ Ongoing EEG in ERPs

➢ Breath, lung, or bowel sounds contaminating the

heart sounds (PCG)
 Example of various types of interference
in biomedical signals

➢ Noise in event-related potentials

➢ High frequency noise in the ECG

➢ Motion artifact in the ECG

➢ Power line interference in ECG signals

➢ Maternal interference in fetal ECG

➢ Muscle-contraction interference in VAG signals

 Noise in ERP (Event-Related Potentials)
➢ An ERP is a signal obtained in response to a
stimulus, The response is of very small amplitude (
10 µv) and is buried in ambient EEG activity and
noise. The waveform of a single response may be
barely recognizable against the background activity
and difficult to analyze
 High frequency noise in the ECG
➢ The noise could be due to the instrumentation
amplifier, the recording system, pickup of ambient
EM signals by the cables etc.
➢The signal is also corrupted by power line
interference at 60 Hz and its harmonics which may
also be considered as a part of high-frequency noise
related to low frequency nature of the ECG signal
 Motion artifact in the ECG
➢ Low frequency artifacts and base-line drift may be caused
in chest-lead ECG signals by coughing or breathing with large
movement of the chest or when an arm or leg is moved in the
case of limb-lead ECG acquisition.

➢ Poor contact and polarization of the electrodes may also

cause low-frequency artifacts.

➢ Base-line drift may sometimes be caused by variations in

temperature and bias in the instrumentation and amplifiers.
Base line drift makes analysis of isoelectricity of the ST
segment difficult.
 Motion artifact in the ECG

➢ A large base-line drift may cause the positive or

negative peaks in the ECG to be clipped by the
amplifiers or the ADC
 Power-line interference in ECG signals
➢ The most commonly encountered periodic
artifacts in biomedical signals is the power line-
interference at 50 Hz.
➢If the power-line waveform is not a pure sinusoid
due to distortions or clipping, harmonics of the
fundamental frequency could also appear.
➢ Power-line interference is difficult to detect
visually in signals having non specific waveform
such as the PCG or EMG.
➢The interference is visible in ECG or carotid pulse.
➢The power spectrum of the signal provide a clear
indication of the presence of Power-line interference
as an impulse or spike at 50 Hz. Harmonics if
 Power-line interference in ECG signals

➢ Fig shows a segment of an ECG signal with 60

Hz interference. The regular or periodic structure of
the interference rides on top of the ECG waves.
 Power-line interference in ECG signals
Fig, shows the power spectrum of the signal. The
periodic
interference is clearly displayed as a spike at not
only its fundamental frequency of 60 Hz but also as
spikes at 180 Hz and 300 Hz, which represents the
third and fifth harmonics respectively
 Maternal interference in fetal ECG

Fig. shows recorded ECG from the woman’s chest

(Maternal ECG)

Fig shows ECG signal recorded from the abdomen of a

pregnant woman (Maternal + Fetus). The abdominal ECG
demonstrates multiple peaks corresponding to the maternal
ECG as well several others at weaker levels and higher
repetition rate. The two signals being combined have almost
the same bandwidth and it is difficult to separate them and
obtained fetal ECG
Muscle-contraction interference in VAG signals
Experiment setup to study the possibility of VMG
(Vibromyography) signals appearing as muscle-contraction
interference in VAG (Vibroarthrography ) signals

➢VAG signals are difficult to analyze as they

have no predefined waveform and it becomes
more difficult to identify any noise or
interference that may be present in VAG signals

➢VAG and VMG signals are nonstationary, also

the bandwidth of the two signals overlap in the
range 0-100 Hz.

➢These factors make removal of the VMG or

muscle-contraction interference from VAG
signals difficult
 Effective DSP Filtering Techniques to remove artifacts
interference in biomedical signals

➢ Time domain filters

1) Synchronized averaging
2) Moving-average filters
3) Derivative-based operators to remove low-frequency artifacts

➢ Frequency-domain filters
1) Removal of high-frequency noise by using butterworth LP filter
2) Removal of low-frequency noise by using butterworth HP filter
3) Removal of periodic artifacts: Notch and comb filters

➢ Optimal Filtering: The Wiener filter

➢ Adaptive filters for removal of interference

 Applications cases

➢ Removal of artifacts in the ECG

➢ Adaptive cancellation of the maternal ECG to obtain the

fetal ECG

➢ Adaptive cancellation of Muscle-contraction interference

in knee-joint vibration signals
 Time-domain filters

Advantages

➢ Spectral characterization of the signal and noise may

not be required

➢ Time domain processing may also be faster than

frequency domain filtering in most of the cases
Time-domain filters-Synchronized averaging
Time-domain filters-Synchronized averaging
 Time-domain filters-Synchronized averaging

Advantages
➢ Spectral characterization of the signal and noise may
not be required
➢ Time domain processing may also be faster than
frequency domain filtering in most of the cases
➢ No spectral content of the signal is lost as is the case
with frequency domain filters

Disadvantage
➢ The realization of the signal that are added for averaging must be
aligned such that the repetitive part of the signal appears at exactly
the same instant in each realization of the signal
➢ If this condition is not met, the waveform of the event in the signal
will be blurred along the time axis
 Time-domain filters-Moving average filters
Segment of ECG signal with high frequency noise is filtered using 8 point MA Filter
Noise level has been reduced, but some noise is still present as attenuation is not
more than -20 dB at most frequencies
Derivative-based operators to remove low-
frequency artifacts
Using Derivative-based operators the base line drift has been
removed but high frequency effects have removed the slow P
and T waves and also altered the QRS complex which can
make the resulting waveform look unlike ECG signal
Application of the filter to the ECG signal with low-frequency noise
is shown below
Low frequency base-line drift has been removed without any significant
distortion of the ECG
 Frequency-domain filters

➢ Filters may be designed in the frequency domain to provide specific

LP, BP, HP or BR characteristics by FIR or IIR design.

➢ Filters may be implemented in software after obtaining FT of i/p

signal
Frequency-domain filters
 Frequency-domain filters
The LP butterworth digital filter with fc= 40 Hz, fs=200 Hz
and N=4 is applied on a carotid pulse signal with high frequency noise
 Frequency-domain filters
The LP butterworth digital filter with fc= 70 Hz,
and N=8 is applied on a noisy ECG signal
Frequency-domain filters
HP Butterworth digital filter with fc= 2 Hz, and N=8, low frequency
artifacts has been removed by the filter whereas the high-frequency noise
is not affected
 Frequency-domain filters

Problem

Design a frequency-domain filter to remove

periodic artifacts such as power-line interference
 frequency-domain filter to remove periodic
artifacts

➢ The simplest method to remove periodic artifacts is to compute the

Fourier transform of the signal, delete undesired components from
spectrum and then compute IFT

➢ Periodic interference may also be removed by notch filters with

zeros on the unit circle in the z-domain at the specific frequency
to be rejected. If fo is the interference frequency, the angles of the
(complex conjugate) zeros required will be , the radius of
the zeros will be unity.

➢ If harmonics are present, multiple zeros will be required at

n represents the orders of all the harmonics present, the filter is then
called a comb filter
ECG signal with 60 Hz interference and after applying
notch filter
 Adaptive filters for removal of
interference
➢ Filters with fixed characteristics are suitable when the characteristics
of the signal and noise are stationary and known

➢ Design of frequency-domain filters require detailed knowledge of the

spectral contents of the signal and noise

➢ These filters are not applicable when the characteristics of the

signal/or noise vary with time (nonstationary). They are also not
suitable when the spectral contents of the signal and interference
overlap significantly.

➢ e.g. When two ECG signals such as those of fetus and the mother
arrive at the recording site and get added in some proportion. The
spectra of the signal in the mixture span the same or similar
frequency ranges and hence fixed filters cannot separate them
 Adaptive filters for removal of
interference
➢ We required a filter that can learn and adapt to the characteristics of
the interference, Estimate the interfering signal, and remove it from
the mixture to obtain the desired signal

➢ This requires the filter to automatically adjust its impulse response,

as the characteristics of the signal and /or noise vary.

➢ Two conditions must be satisfied

(1) The filter should be adaptive
(2) The filter should be optimal
 Adaptive noise canceller

➢ Where v(n) is signal of interest, m(n) is primary noise, x(n) is the

primary observed signal, y(n) is noise estimate
➢ Assume that v(n) and m(n) are uncorrelated
➢ Adaptive filtering requires a second i/p known as the ‘reference i/p’
r(n) that is uncorrelated with the signal of interest v(n) but closely
related to or correlated with the interference or noise m(n) in some
manner
 Application: Removal of artifacts in the ECG

➢ Design filters to remove base-line drift, high frequency noise and

power line interference in ECG signals

Solution: Take the power spectrum of the given ECG signal.

High amount of spectral energy present near DC, from 100 Hz to 500 Hz,
and at power line frequency and its harmonics located at 60 Hz, 180 Hz,
300 Hz, and 420 Hz.
 Application: Removal of artifacts in the ECG

A butterworth LP filter with order N=8, and Fc= 70 Hz, a Butterworth High pass filter of
order N=8 and Fc=2 Hz and a comb filter with zeros at 60 Hz, 180 Hz, 300 Hz and 420 Hz
were applied in series to the signal.
After Lowpass filter

➢The base-line drift is present in the output of the low pass filter and
power-line interference is present in the output of the LP and HP filters.
The output after comb filter is free from interference.
 Application: Adaptive cancellation of the maternal ECG
to obtain the fetal ECG

Widrow et al. describe a multiple-reference ANC for removal of the maternal ECG in order to
obtain the fetal ECG. The combined ECG was obtained from a single abdominal lead,
whereas the maternal ECG was obtained via four chest leads.

➢ The signals were pre-filtered to the bandwidth 3-35 Hz and a sampling rate of 256 Hz
was used. The adaptive cancellation has resulted in the signal shown below
 Application: Adaptive cancellation of muscle-contraction
interference in knee-joint vibration signals
Fig. below shows plot of the VAG signal of a patient and simultaneously recorded channel
of muscle-contraction interference. The results of adaptive filtering of the VAG signal with the muscle-
contraction interference channel as the reference is shown in third waveform for the result using LMS
filtering and fourth waveform using RLS filtering. A single stage LMS filter with variable step size µ with
M=7, µ=0.05, and α=0.978. The RLS filter use M=7 and λ=0.98