now we come to the second part of smart cities
and smart homes in part one we spoke about the need for smart cities the challenges in
building smart cities and few of the different issues with respect to building smart cities
so in this particular lecture again we will be focusing on smart cities and these again
you know what ever we will be discussing on smart cities are also applicable for smart
homes as well but we will have you know a focus on smart of home in another lecture
but here we are going to discuss some of the technical issues behind enabling smart cities
some of the technical issues so let us consider something before we go
further we have already seen that we have all these different components the transport
railways schools and etcetera etcetera and let us say in between we have this population
population means the citizens right so all of these things these different components
let us say this is transport so transport component has different sensors has different
actuators these this is let us say railway railways all source has the same other components
schools hospitals they all have different sensors and different other iot devices which
generate which generate data which generate and data and this data has different ah you
know characteristics with respect to volume this is gigantic you know huge volumes of
data that each of these different components they are generating coming at high velocities
has different you know different types of data media data you know multimedia data text
data and so on and so on and so forth they are at a these different characteristics of
big three vs to five vs to seven vs and this is something that we spoke about in a previous
lecture already so i not going to elaborate on these but what
is required is to do some good planning so what are you going to do with all this data
so you know one possibility is that these data can be made available to the population
to the citizens but making the data available just like that will not help so you have to
do some processing so let us say that that processing is also done then you have to you
have to fuse these data that are made available from these different sources let us say transportation
data and health care data sense from different points made available from these different
locations has to be this data have to be fused together in order to give better insight about
different things in a smart city so that part is really really challenging so one thing is to deal with this kind of
data you know this big data that is coming in real time analyzing you know cleaning up
processing analyzing this kind of data in real time this is one thing but in addition
you have to fuse the data that are cross to that are coming from different sources and
that is a highly challenging issue and like this there are different other issues in the
building of smart cities in the previous lecture we spoke about the overall idea the philosophy
behind smart cities but then you have to make it technically made possible it is not like
you know you connect few sensors and then you know communication will be zigbee wifi
etcetera and then you make the data available no that data is going to be of limited use
to the corresponding users of the stake holders so that is going to be of limited use so you have to you have to do some better
job by fusing the data together and then making that kind of fused data which has more in
sight which will give more insight ah you know that will be more useful so let us look
ahead and see what we have for us in data fusion so data fusion basically you know you
are talking about in a smart city environment we are talking about enormous volumes of data
that are produced periodically and the challenges include making these particular type of data
available and so that the incoming larger data can make more sense can make more sense
and with the help of this data from these different sources the large volumes of data
from this different sources different predictions different analytics should be should be executed
so the quality of data precession and the accuracy basically affects the quality of
decision making in this kind of iot based smart city environments so data fusion basically
enables optimum utilization of this massively collected data from different sources across
different platforms so multi sensor data fusion is very important
which basically combines information from multiple sensor sources it enhances the ability
of the decision making system to include a multitude of variables prior to arriving at
a decision this is what i was telling you so it is not like clubbing two data together
alone but you considered the different you know different different issues and considered
the different variables that are affecting ah ah ah these systems and you know taking
all of these into account not just the sensor data taking all of these multitudes of ah
you know variables effecting the data ah and affecting the system together you come with
you arrive at a decision and make that decision made available not just decision but may be
different options and make that decision or the different options available to the users so data fusion basically will help you in
doing this inferences are drawn from multiple sensor type data and ah these are typically
ah you know ah ah qualitative the inferences are qualitative and ah ah you know these basically
are of more insight these are more insightful these are more meaning full ah ah than the
single sensor type data so these putting these different types of it ah you know different
types of data together and trying some kind of you know arriving at some kind of intelligent
decision that is more insightful than the individual data information fusion generated
from multiple heterogeneous sources provides for better understanding and understanding
of the operational surroundings the challenges include we are dealing with
an environment where the data has lot of imperfection imperfection due to inherent devices devices
like sensors etcetera where there is lot of uncertainty around the environment there are
lot of inaccuracies that can keep in so there is lot of inaccuracy uncertainty in the data
and that basically leads to imperfections ambiguity is another we are talking about
an environment where there are you know data that that that are collected have lot of different
outlets outlet means that there would be some data points which will be far away from the
similar data points in the cluster and there could be some missing data as well so a ambiguity in the data can also creep
similarity there can be conflicts in data data that are connected from different sensors
about the same thing they might be conflicting they might be contradicting alignment is like
this that it arises when the sensor data frames are converted to a singular frame prior to
transmission so that also has to be done you know so that alignment you know into singular
frame that is challenging different other trivial features for example processing of
trivial data features may bring down the accuracy of the whole system and ah these are some
of the challenges that have to be talked on when you are talking about data fusion in
an iot environment so what are the opportunities so collective
data is reaching information and it generates better intelligence better insight compared
to the single source data from different individual sensors so putting these data together you
know will make you to get better insight so you know what is the required is to optimally
amalgate optimally amalgate means that integrate optimally integrate the data because you know
the more and more you integrate you know it is possible to get more insight but at the
same time you know that also has to be done in real time to be you know for that decision
to be more meaning so so optimal amalgation of ah amalgamation
of data then enhancing the collective information content obtained from multiple low power low
precision sensors and enabling ah ah data data fusion basically ah enables the hiding
of critical data sources and the semantics and that is useful for military applications
for medical use cases the different stages of data fusion include decision level which
is basically an ah you know talking about an ensemble coming up with an ensemble of
decisions then feature level you know that means that the different features you know
you fuse with respect to the different future ah features at the feature level the integration
is done the fusion is done so it is basically fusion of information prior
to decision making and pixel level is fusion of information at the imaging device level
itself ok so at the imaging devices that fusion is done in the device itself and single level
basically fusion of information at the sensor node or within the local area network itself
the mathematical methods of data fusion include using probability based schemes such as bayesian
analysis statistics recursive methods ei based schemes such as artificial neural network
machine learning algorithms deep neural networks convolutional neural networks theory of evidence
based ah ah ah you know evidence based schemes for example ah belief functions taking use
of belief functions transferable belief models so these are the different mathematical methods
that are used in order to come up with these intelligence from the different data that
are ah you know that are secured from the different iot devices so ai artificial intelligence comes as a big
helper in enabling this so you know let us consider this particular figure so traditionally
what happens is you have these different sensors and the sensor data has to be transmitted
over the communication medium and has to be you know based on that some actuation is going
to happen but how how that actuation is going to be made ah you know made possible is it
from one or two of these sensors in a based on these ah sensor value you are going to
actuate or can we do something better so for betterness betterment what can be done is
some kind of decision making has to be done with the help of intelligence by by adding
intelligence between these different sensors and the actuators we can make things better
make things improved so how is that made possible with the help of artificial intelligence tools
methods algorithms and so on so ai come as a rescuer over here and what
ai can do is it can make highly accurate decision making possible between the sensors and the
actuators so let us consider the scenario of decision fusion for autonomous vehicles
so for autonomous vehicles like autonomous cars etcetera etcetera they interact a lot
with the environment you know when there is a driver less car for instance they need to
take help of this different sensors they need to also communicate with ah ah you know the
with the satellite with the gps with the help of technologies such as you know lidar technology
for obstruction you know for getting a map of the obstructions ahead or the ultrasonic
sensors can help in even checking some different obstructions that are ahead of them in a small
scale lidar can give a bit bigger picture whereas ultrasonic sensors can give a small
scale picture of what is ahead of the autonomous vehicle so autonomous cars you know they basically
are collecting different data from different sources through different technologies like
lidar sensor sensor networks ah you know from satellites through gps ah and different cameras
ahead of you know in front of them so they all these different data of different types
as you can see are connected and they are sent to the server now you know these data
of different types you know individually they do not make much sense they are of limited
help but together can these data be fused together so that the car the autonomous car
can get some kind of decision making about how it is going to proceed or whether it is
going to turn left or right or what it is going to do if it if it sees some pedestrians
in front then what it is going to do like this kind of thing is made possible with the
help of data fusion ah ah data fusion technology so all these decision making ah through you
know of the data that is connected from the different sources you know that is basically
made possible with the help of data fusion smart parking i already spoke about smart
parking in a in the previous lecture but let us dig into this smart parking ah ah little
bit further smart parking is very much an important component now a days we have smart
parking solutions in different cities already started to be deployed so ah in a smart parking
environment what happens is you know the user knows ahead of actually going to that particular
spot that which of the spots in the cities have free parking spots right which of these
different parking lots have free parking spots and then accordingly the driver can make a
decision about where to go and park the car so smart parking basically shortens the parking
search ah parking search time of the drivers so basically you know searching for the different
parking lots that search time would be ah reduced will be shortened and you know it
the parking is going to be made efficient it reduces the traffic congestion reduces
the pollution by keeping unnecessarily lingering vehicles off the road so you know so what
would happen is in a smart way you know where to go and where to park that way it is not
going to happen that you are in a queue waiting for your parking your engine is on you are
polluting the environment so in a smart parking basically also helps in reducing pollution
unnecessarily ah in a city it reduces the fuel consumption and costs as well and these
are all actually interlinked so that ah you know fuel consumption more fuel consumption
more pollution more costs are involved ah you know so like this these are all interlinked
increases the urban mobility and the shorter parkings search time results in more parked
time and hence more revenue so in a smart parking scenario we are talking
about ah you know information collection system deployment rather i would start with the system
deployment system is deployed information is collected and the surfaces are disseminated
to the end users so these are the different functional layers of smart parking in terms
of information collection information is collected from the sensors the individual sensors in
the car in the parking lot there are different parking meters the sensors are networked together
so you have the sensor network and also the crowd sensing crowd sensing basically is from
the crowd the from the different sensors in the mobile force ah in the smart phones for
instance you are able to collect the different data and these data will help in in decision
making so all these data taking together and fusion
of these data will help in decision making then we have the system deployment ah with
respect to the software system that has to be developed the information management of
the data the e parking ah you know guidance system that will help in guiding the vehicle
about where to go how to go and ah you know ah and parking the car there ah then the data
analytics over all so these are the different system level deployment issues in smart parking
service dissemination with respect to dynamic pricing strategizing infrastructure based
information infrastructure free information so infrastructure based and infrastructure
free so infrastructure free from the different sensors you know these are not connected to
the regular infrastructure like wifi etcetera etcetera this is infrastructure free infrastructure
based means like from the regular internet infrastructure from the regular city communication
infrastructure like wifi and ah ah you know ah ah like three g four g you know the cellular
ah networks so on these are all like the infrastructure based and then infrastructure free is what
i just told you with the help of sensors ad hoc networks formed out of these different
mobile device of different users etcetera then parking choice and vehicular activities
these all contribute to the building of ah ah services required for smart ah smart parking
in terms of information sensing in smart parking the sensing can be done from stationary sensors
or from mobile sensors stationary sensors like ah you know if you are collecting the
data from stationary sensors you need large number of sensors to be deployed at different
points which will detect the presence or absence of ah different vehicles or from mobile sensors
where fewer sensors would be required compared to the case of stationary sensors and these
mobile sensors the fewer mobile sensors would collect information along the root when they
go by energy management in smart cities you need
energy solutions so energy efficient solutions it is required to light weight the protocols
because you are dealing with a highly resource constraint environment and at the same time
energy consumption has to be reduced for you know reasons of greenness environment ah ah
and so on so light weight protocols are required ah it is required to schedule ah the optimization
of ah ah you know optimization of energy consumption and ah then predicting models for energy consumption
is another ah important thing then you have the cloud based approach low power conceivers
cognitive management framework these are the different energy efficient solutions for for
energy management in smart cities energy harvesting solutions would include
technique help of ah these ah you know harnessing energy from these renewable sources of energy
such as sun wind heat vibration rf sources now a days people are talking about harvesting
energy from radio frequencies as well so from rf sources harvesting energy from sun wind
heat vibration and like this so many different types of ah you know ah sources of energy
ambient sources of energy are there and how you can harness the energy from all of these
different sources energy harvesting solutions would include
dedicated energy harvesting ah by the deploying different ah ah you know different ah ah sources
like like solar panels etcetera you know deployed pre deployed energy sources are intentionally
deployed near the iot sources to power these iot devices for example in our agricultural
field closer to the sensor node very close to the sensor node we have these solar panels
and these panel solar panels basically power the sensor nodes that we have deployed in
our agricultural field ah ah and these sensor nodes are basically deploy ah you know were
developed in the swan lab of our ah institute so ah the distance between the device and
the source ah the sensitivity of the harvesting circuit and the environment these basically
ah ah are contributors to determining the amount of energy that is harvested
so with this we come to an end of the second part of the lecture on ah on smart cities
here we have mostly covered issues such as how to handle the data that is received from
this different sources we can try to make inferencing with the help of these standalone
sensor data that are received the separate individual data that are received or is it
possible to do better you know it is possible to do better by fusing the data from the different
sources together with the help of intelligence and so on so ah this is the end of the smart
cities part two the ah the next in the next part we are going to talk about few other
different issues of building ah these smart environments and there the focus will be on
smart homes thank you

Woodworking Plans Banner

You May Also Like