real time multisensor technique for false ......2018/06/02 · building automation infrastructure,...

Journal of Analysis and Computation (JAC) (An International Peer Reviewed Journal), www.ijaconline.com, ISSN 0973-2861

Volume XI, Issue I, Jan-June 2018

Hetusha Patel1, Suhani Patel2, Prof. Meet Shah3, Dr. Lokesh Sharma4 1

REAL TIME MULTISENSOR TECHNIQUE FOR FALSE ALARM

REDUCTION USING IOT

Hetusha Patel1, Suhani Patel2, Prof. Meet Shah3, Dr. Lokesh Sharma4

[email protected] , [email protected], [email protected], [email protected]

1Rollwala Department of Computer Science, Gujarat University, Ahmedabad.

ABSTRACT:

The present development number of worldwide systems administration enormously affects the participation of shrewd components, of discretionary kind and reason that can be found anyplace and cooperate with each other as indicated by the predefined convention. Besides, these components must be shrewdly coordinated with a specific end goal to help disseminated detecting as well as checking/control of true wonders. That is the reason the Internet of Things (IoT) idea raises like another, promising worldview for Future Internet advancement. Considering that Wireless Sensor Networks (WSNs) are imagined as basic piece of subjective IoTs, and the conceivably immense number of collaborating IoTs that are normally utilized as a part of this present reality wonders checking and administration, the unwavering quality of individual sensor hubs and the general system execution observing and change are certainly testing issues. A standout amongst the most intriguing true marvels that can be observed by WSN is indoor or open air fire. The joining of delicate processing advancements, as fluffy rationale, in sensor hubs must be explored keeping in mind the end goal to pick up the reasonable system execution observing/control and the maximal expansion of segments life cycle. Numerous viewpoints, for example, courses, channel get to, finding, vitality effectiveness, scope, arrange limit, information collection and Quality of Services (QoS) have been investigated broadly. There are two fluffy rationale approaches, with transient attributes, are proposed for observing and deciding certainty of flame keeping in mind the end goal to advance and lessen the quantity of standards that must be checked to settle on the right choices thus we are going contribute our idea Smoke Sensor, Flame Sensor, Moisture Sensor. We accept that this decrease may bring down sensor exercises without applicable effect on nature of activity and diminish the likelihood of false alert contributing the proficiency, power and cost viability of detecting keeping in mind the end goal to get an ongoing check of proposed approaches a model Sensor web hub, in view of Representational State Transfer (RESTful) administrations, is made as a foundation that backings quick basic occasion flagging and remote access to sensor information through the Internet A probabilistic neural system (PNN) is a sustain forward neural system, which is generally utilized as a part of arrangement and example acknowledgment issues And furthermore used to diminish FAR.

Keywords: Temperature sensor, wireless sensor networks, fuzzy logic, Moisture sensor, Smoke sensor, Flame Sensor, False Alarm reduction,PNN.

[1]INTRODUCTION

Low cost Wireless Sensor Networks (WSNs) have been developed and widely used in a variety

of applications like automotive engineering, defense, safety, medical, home and environment [1],

precision agriculture, animal tracing, security, surveillance, urban terrain tracing, civil structure

monitoring, transportation and entertainment. Some environmental applications of WSNs involve

tracking the movements of animals and people, monitoring environmental circumstances that

affect crops, chemical biological detection, precision agriculture, forest fire detection,

meteorological or geophysical research and flood detection [2]. WSNs are mainly designed for

monitoring and controlling events in a high risk area such as forests, mountainous terrain and

cross-border; therefore, frequent maintenance becomes costly and unrealistic. As a result, the

proposed system introduces a method that is self-sufficient in providing a stable power supply.

http://www.ijaconline.com/




Solar systems can provide the power needed for a WSN and minimize human intervention by

creating completely autonomous operation. Numerous techniques for making an intelligent

decision With the threats, risks and dangers that exist today, fire and particularly fire security

systems play an increasingly significant role in life-safety operations. Fire can cause massive

damage to the indoor or outdoor area, creating the severe life threatening conditions, making the

early and accurate residential fire detection extremely important for prompt extinguishing,

reducing the damages and potential life losses. The integration of fire-safety systems within

building automation infrastructure, through fast delivery of sensed data, quick response, access

control, video surveillance, fire detection and alarm, and emergency communications aides the

effective incident management. One of the ways to monitor and detect fire is to use the Wireless

Sensor Networks (WSN) composed of low-resourced sensor nodes. WSNs are manufactured and

deploy for different purposes by various group. They are composed of spatially distributed nodes

equipped with sensing devices (to monitor environmental conditions at different location),

processing unit, communication components (wireless transmitter/receiver), storage unit, and an

energy source [3]

Fig 1: Typical architecture of a sensor node

Signal detection has long been based on statistical-decision theory. When a distorted signal

arrives at the receiver, the receiver would require to detect correctly the presence of such a signal.

Traditionally in making a signal detection decision, binary logic with only two elements i.e. {1,

0) is assigned to the signal and the decision is then optimized using certain decision functions .

This method of detection decision is based on an a priori signal distribution and the decision

functions are derived from the statistical spectrum of the disturbances. For example, in radar

detection the incoming signal is first threshold and then processed by a decision function using

the M-out-of-N or CFAR (Constant False Alarm Rate) algorithm. When the signals being

processed are in binary logic, they are quantized functions resulting in incomplete information.

In fact, many complicated but refined detection algorithms have claimed to be optimal. They are

only optimal in the sense that their performance is maximized with the incomplete information

being processed. With the present development in fuzzy logic, signal detection is not only

restricted to the likely hood of an event to probability, but can be extended to the degree to which

an event occurs.

Historically, the utilization of sensor data in engineering control systems involves several

tasks that are performed with a human operator somewhere in the decision loop. These tasks

include data monitoring to determine if data should be processed, as well as monitoring system

output to determine whether the system is performing as expected. The data produced by sensors

are quantitative, and great care and expense is taken to produce data with good precision and

integrity. Control systems evaluate the data based on a set of rules learned by the human expert,

usually from experience. Unless artificially derived by a professional knowledge engineer, these

rules are not “crisp,” i.e., the words used are open to human interpretation and require the

application of common sense and judgment to be successfully applied in a decision-making role.

A rule contains deterministic conditions that are actually fuzzy in nature and are interpreted by





the operator during operations. Such rules are appropriate for fuzzy set modeling, where decisions

made by fuzzy logic can be as good as those made by the expert [4], [5], [6], [7], [8], [9], [10].

This differs from modem control theory. Let’s take PID control, for example. The letters PID

stand for Proportional, Integral, and Derivative. The P represents a term in the controlling

equation that is proportional to the offset from the set value; the

I represents a term that is the integral of the offset; and the D stands for a term that is the

derivative of the offset. Modern control uses mathematical models, and if a highly accurate model

can be made, it is theoretically a very precise method. In most real control systems, it is both

difficult and time-consuming to create an accurate mathematical model. This makes modern

control inconvenient for certain applications. It is not uncommon for control systems to assume

linear relationships, and then overcome inaccuracies by adjusting parameters. A FL control

system has no analytical models of the process to be controlled. It knows no equations that

describe the physical system. In a well-known pendulum balancing system, for example, only

two variables describing the system’s state at a particular time step are known. Yet the system

maintains excellent control.

Human operators who work with modern controllers develop skills to help them

compensate for controller errors. These skills are based on cognitive models or patterns of how

the controllers work and are not mathematical, because human operators do not use calculations.

Clearly, there is a difference in the way human experts perform control tasks. Human senses are

fuzzy senses. Fuzzy logic does a better job of representing what is actually in the mind of the

human expert, because it doesn’t impose artificial structures on the expert’s knowledge.

Descriptions using deterministic rules introduce an unnecessary transformation of expert

knowledge. Fuzzy systems are unique in that they attempt to emulate what a human does, not

what a human says is done. Humans infer the process from observing the results. Using sensor

inputs and a model of the process, machines tend to predict the results. Therefore, fuzzy logic

control requires innovative sensors-sensors that are results-oriented, not process-oriented. Fuzzy

logic controllers are relatively insensitive to variations in the parameters of the system

environment. Fuzzy systems directly encode structured knowledge, but do so in a numerical

framework. Fuzzy systems allow users to articulate rules linguistically by defining fuzzy sets as

trapezoids or triangles [11]. Once the variables and fuzzy sets have been identified and defined,

a prototype fuzzy system can be created in minutes.

[2] LITERATURE REVIEW

The WSNs are typically used to monitor some of the parameters of environmental

processes which are complex, ambiguous and vagueness embedded in their nature. Most

previously performed researches in WSNs rely on precise, also called crisp, values to specify the

parameters of interest. As the consequence of such a rigid approach, the sensor readings can be

imprecise and unreliable, making the using of crisp values to describe WSN parameters

inadequate. In spite of the fact that sensor nodes have highly constrained resources

(microcontroller, memory, battery, communications), numerous new functionalities have been

proposed for WSNs. Considering that WSNs are envisioned as an integral part of the Future

Internet, supporting its extension to the physical world, the incorporation of soft computing (SC)

technologies in sensor nodes may lead to potential network performance improvements, since it

provides effective parameter combination and can be directly executed by the sensor nodes.

Integration of soft computing technologies (fuzzy logic, neural networks, fuzzy rule-based

systems, data mining techniques, etc.) in sensor nodes is a good example of an application adapted

to WSNs [12]. Authors of [13] believe that crisp values cannot adequately handle the often





imprecise sensor readings and that is why they propose fuzzy values instead of crisp ones claiming

that they significantly improve the accuracy of event detection. Therefore, WSN powered by a

fuzzy logic detection mechanism, behaves like intelligent and power efficient sensing network.

The incorporation of fuzzy logic or fuzzy approach in WSNs is presented in numerous papers.

Fuzzy logic provide a straightforward way to arrive at a definite result based upon vague,

ambiguous, inexact, noisy, or missing effort data or the information [14]. What makes fuzzy logic

suitable for use in WSNs is that: it can tolerate unreliable and imprecise sensor readings; it is

much closer to the human way of thinking than crisp logic; and it is much more intuitive and

easier to use, when compared to other classification algorithms based on the probability theory.

In WSNs, fuzzy logic has been generally used to improve the decision-making process, reduce

resource consumption, and increase performance. A main difficulty of using fuzzy logic is that

the number of rules grows exponentially to the amount of variables (with n variables, each of

which can take m values, the number of rules in the rule-base is mn ) demanding a significant

amount of memory for storing the rule-base [14]. To solve this problem, efficient rule- 66 Mirjana

Maksimović et al. base reduction techniques, with different advantages and modeling needs, are

developed. A key property of these techniques is that they do not affect the application accuracy.

Some of the rule-base size reduction techniques are described in [12-15] As Nikolay

Brayanov,etal proposed Recently fuzzy logic has increasing its popularity. One of the main

reasons is that it gives abstraction, so a system could be controlled, even if it is not fully

describable. On the other hand, this approach simplify’s incorporation of data from multi sources

and its right usage. The paper is focused on a grown-up area of temperature control, so the data

is homogenous and simple. This applicability demonstration is not limitation and the approach

could be used for any type and quantity of data.. In this type of products that consist of a network

of sensors, it is beneficial to use all available data. The result of this research is analyzed,

demonstrating better accuracy as result of a fusion of sensors[15] Internet of things is an

interconnection of physical devices embedded with electronics, software, sensor which is capable

of collecting data from the surrounding and sending data over internet is called IOT. The fire

detection gathers all of the techniques and processes that contribute to early detection of a fire.

We identify three main categories: Smoke detection, Flame detection and Temperature detection.

Automatic fire alarm system provides real-time surveillance, monitoring and automatic alarm.

An automatic fire alarm system based on WSN is developed, which is intended for high-rise

buildings. To provide early extinguishing of a fire disaster, large numbers of detectors which

periodically measure smoke attention or temperature are deploy in buildings.

As Proposed by Md Iftekharul Mobin,etal Safe From Fire (SFF) is an intelligent self

controlled smart fire extinguisher system assembled with multiple sensors, actuators and operated

by micro-controller unit (MCU). It takes input signals from various sensors placed in different

position of the monitored area, and combines integrated fuzzy logic to identify fire breakout

locations and severity. Data fusion algorithm facilitates the system to discard deceptive fire

situations such as: cigarette smoke, welding etc. During the fire hazard SFF notifies the fire

service and others by text messages and telephone calls. Along with ringing fire alarm it

announces the fire affected locations and severity.[16] To prevent fire from spreading it breaks

electric circuits of the affected area, releases the extinguishing gas pointing to the exact fire

locations. This paper presents how this system is built, components, and connection diagram and

implementation logic. Overall performance is evaluated through experimental tests by creating

real time fire hazard prototype scenarios to investigate reliability. It is observed that SFF system

demonstrated its efficiency most of the cases perfectly





As said by P. Singhala The aim of the temperature control is to heat the system up to

delimitated temperature, afterward hold it at that temperature in insured manner. Fuzzy Logic

Controller (FLC) is best way in which this type of precision control can be accomplished by

controller. During past twenty years significant amount of research using fuzzy logic has done in

this field of control of non-linear dynamical system[17] Here we have developed temperature

control system using fuzzy logic. Control theory techniques are the root from which convention

controllers are deducted. The desired response of the output can be guaranteed by the feedback

controller

As Proposed by Mohammed Elnour A/Alla an artificial intelligent control method for

temperature control system and is suitable for low temperature applications such as laboratory

equipments (e.g. ovens and incubators). The proposed design uses fuzzy logic as a control method

that maintains the temperature of simulated heater to the desired point. Microcontroller based

circuit is built to acquire data from sensor, actuate heat element and communicate with computer

workstation. MATLAB fuzzy logic controller is designed, tested, and tuned to control the circuit.

The Fuzzy Logic Controller performance is evaluated in several situations by comparing it with

conventional Proportional Integral Derivative (PID) controller in terms of speed of response to

the desired setting value, overshoot in fixed set point and robustness against disturbance. FLC is

fast in response to the setting with compare to PID, and more stable against external disturbance.

Both of FLC and PID have neglected overshoot value and steady state error, but FLC has

noticeable deviation in high set points.

As prospoed by Yudhajit Das,etal a room temperature and humidity controller using fuzzy

logic. The proposed model consists of two fuzzy logic controllers to control temperature and

humidity respectively. The first controller accepts two input values- the current temperature as

detected by temperature sensor and its deviation from user set-temperature, and controls the speed

of heat-fan and cool-fan accordingly . When the current temperature in the room reaches set point,

it serves as one of the input for second fuzzy logic controller that controls the humidity. The ideal

relative humidity level for user’s set temperature is preset in the system. Current humidity in %

as detected by the humidity sensor in the room serves as the second input to the controller. The

humidifier and exhaust fan speed is controlled accordingly to maintain the correct humidity level

for that temperature. This research work will increase the capability of fuzzy logic control systems

in process automation with potential benefits. MATLAB-simulation is used to achieve the

designed goal.

As proposed by Koushik Anand,etal An intelligent drip irrigation system optimizes water

and fertilizer use for agricultural crops using wireless sensors and fuzzy logic. The wireless sensor

networks consists of many sensor nodes, hub and control unit. The sensor collects real-time data

such as temperature, soil humidity. This data is sent to the hub using the wireless. Real Time

MultiSensor Technique for False Alarm Reduction using IOT technology. The hub processes the

data using fuzzy logic and decides the time duration for keeping the valves open. Accordingly,

the drip irrigation system is implemented for a particular amount of time. The whole system is

powered by photovoltaic cells and has a communication link which allows the system to be

monitored, controlled, and scheduled through cellular text messages. The system can quickly and

accurately calculate water demand amount of crops, which can provide a scientific basis for

water-saving irrigation, as well as a method to optimize the amount of fertilizer used.





As described Suhas Sayajirao Jahdav,etal [18] KITCHEN environment monitoring is one

of the important measures to be closely monitored in real-time for safety, security and comfort of

people. With the advancements in Internet technologies and Wireless Sensor Networks (WSN),

a new trend in the era of ubiquity is being realized. Enormous increase in users of Internet and

modifications on the internetworking technologies enable networking of everyday. The paper

proposes a Raspberry pi based kitchen monitoring system. Raspberry Pi is used as a Embedded

Web Server, User can control Set of devices from Phone/PC Web Browser. We have designed

and implemented a compact wireless sensor network with internet capability. The system can

monitor the status of kitchen and send an alert SMS via GSM network automatically to users. The

system has the capability to control through internet, where the subject of received email is read

by the developed algorithm fed into Raspberry pi and then the system responds to the

corresponding instruction with high security. The user can directly log in and interact with the

embedded device in real time without the need to maintain an additional server. Here the project

also proposes the sensors interfacing with the controller and GSM modem too. If there is the Gas

detection or fire the message will be sent through the GSM. The system is modularly built,

allowing different modules to be added. In addition, it is flexible to accommodate a wide range

of measurement devices with appropriate interfaces. It has a variety of features such as energy

efficient, intelligence, low cost, portability and high performance As Described by Meana-Llori

,etal The Internet of Things is arriving to our homes or cities through fields already known like

Smart Homes, Smart Cities, or Smart Towns. The monitoring of environmental conditions of

cities can help to adapt the indoor locations of the cities in order to be more comfortable for people

who stay there. A way to improve the indoor conditions is an efficient temperature control,

however, it depends on many factors like the different combinations of outdoor temperature and

humidity. Therefore, adjusting the indoor temperature is not setting a value according to other

value. There are many more factors to take into consideration, hence the traditional logic based

in binary states cannot be used. Many problems cannot be solved with a set of binary solutions

and we need a new way of development. Fuzzy logic is able to interpret many states, more than

two states, giving to computers the capacity to react in a similar way to people. In this paper we

will propose a new approach to control the temperature using the Internet of Things together its

platforms and fuzzy logic regarding not only the indoor temperature but also the outdoor

temperature and humidity in order to save energy and to set a more comfortable environment for

their users Finally, we will conclude that the fuzzy approach allows us to achieve an energy saving

around 40% and thus, save money.

As described by Qais Qassim[19] Pervasive and sustained cyber attacks against information

systems continue to pose a potentially devastating impact. Security of information systems and

the networks that connect them is becoming increasingly significant nowadays than before as the

number of security incidents steadily climbs. The traditional ways of protection with firewall and

encryption software are no longer sufficient and effective. In this struggle to secure the data and

the systems on which it is stored, Intrusion Detection and Prevention System (IDPS) can prove

to be an invaluable tool. IDPS can also, be a very useful tool for recording forensic evidence that

may be used in legal proceeding. The intrusion detection and prevention system have provided a

high detection rate in detecting attack attempts. However, IDPS performance is hindered by the

high false alarm rates it produces. This is a serious concern in information security because every

false alarm can onset a severe impact to the system such as the disruption of information

availability because of IDPS blockage in suspecting the information to be an attack attempt. The

aim of this paper is to propose a strategy to reduce these false alarm rates to an acceptable level





to maintain the total security against serious attacks by implementing a fuzzy logic-risk analysis

technique for analyzing the generated alarms.

As Described by Susan L. Rose-Pehrsson[20] The Navy program, Damage Control Automation

for Reduced Manning DC-ARM , is focused on enhancing automation of ship Ž . functions and

damage control systems. A key element to this objective is the improvement of current fire

detection systems. As in many applications, it is desired to increase detection sensitivity and,

more importantly, increase the reliability of the detection system through improved nuisance

alarm immunity. Improved reliability is needed such that fire detection systems can automatically

control fire suppression systems. The use of multi-criteria-based detection technology continues

to offer the most promising means to achieve both improved sensitivity to real fires and reduced

susceptibility to nuisance alarm sources. A multi-signature early warning fire detection system is

being developed to provide reliable warning of actual fire conditions in less time with fewer

nuisance alarms than can be achieved with commercially available smoke detection systems. In

this study, a large database consisting of the responses of 20 sensors to several different types of

fires and nuisance sources was generated and analyzed using a variety of multivariate methods.

Three data matrices were developed at discrete times corresponding to the different alarm levels

of a conventional photoelectric smoke detector. The alarm times represent 0.82%, 1.63% and

11% obscurations per meter. The datasets were organized into three classes representing the

sensor responses for baseline non-fire , fires and nuisance sources. A robust data analysis strategy

for use with a sensor array consisting Ž . of four to five sensors for early fire detection and

nuisance source rejection was developed using a probabilistic neural network PNN Ž . that was

developed at the Naval Research Laboratory for chemical sensor arrays. The analysis algorithms

described in this paper evaluate discrete samples and develop classification models that examine

individual chemical signatures at discrete points. Published by Elsevier Science S.A.

[3] PROPOSED SYSTEM

This section introduces the principle of a new sensor fire detection algorithm. According

to the current situation in automatic fire detection in technology, two restrictions were imposed on the design of the new detection algorithm

1.It should be possible to implement the algorithm on a common 8 bit microprocessor systen1 already used in existing detectors.

2. With respect to the environmental regulations on ionization sensors the new algorithm should monitor optical smoke density and temperature serving as a possible replacement for

ionization detectors.

The capabilities of a detection system mostly depend on the signal processing applied.

This signal processing can be divided into two subsystems: signal preprocessing (i.e. analog and

digital filtering, statistical processing) and a decision network. In a simple threshold detector these

subsystems are the analog amplifier and the threshold device, respectively. The decision network

of the new detection algorithm is realized by a fuzzy expert system. Figure 2 illustrate the

structure of a multi-sensor detector based on the new detection algorithm. The following

subsection will briefly introduce the basic concepts of to lie fuzzy set theory and then the

essentials of a fire detection algorithm using fuzzy logic and taking into account the design

restrictions will he presented.





Fig 2. Hybrid Structure

Very promising solution module for hardware implementation in fire detection and control

has been developed by using new algorithm. New algorithm is combination of fuzzy logic and

PNN. Fuzzy AHP and PNN both are used to improve accuracy of Fire detection system. Fuzzy

AHP will be replaced by PNN. Implement fuzzy logic with PNN for fire detection. Fuzzy logic

can be used for detecting fire and find location of fire and PNN gives better accuracy. The

combination of fuzzy logic and PNN improve the accuracy of fire detection, reducing the false

alarm and make fire detection system more intelligent.

Fig 3. Structure of a multi sensor detector

We are going to use Raspberry PI. To which there will be four sensor are connected. Temperature,

moisture, smoke, flame It will be connected to PC where the sensor reads the status signal. Where

it’s applied to fuzzy logic and decision control comes into play. Each and every sensor we are

going to apply the validate FAR. Which will be an key factor to provide advancement over the

previous concept. Each an every sensor will be provided with certain kind of threshold ,which

will be able to prevent the false alarm

Live Aftermath Fan will be turned on when the room temperature is 40. If the temperature keeps

on fluctuating (39,40) then he fan will be turned off/on. We will be providing certain kind of

threshold to prevent the false alarm. It means temperature will be taken into account after each 10 seconds.

[4] FUZZY SET THEORY





Fuzzy Logic resembles the human decision-making methodology. It deals with vague and

imprecise information. This is gross oversimplification of the real-world problems and based on degrees of truth rather than usual true/false or 1/0 like Boolean logic.

Traditional Set Theory: A ∩ A = 0

Fuzzy Set Theory: A ∩ Ā ≠ 0

One assigns non-binary membership, or degrees of membership, to classes of events (fuzzification).

Signal Detection Theory

Mixed Implication Functions

H = min (s,r)

M= max (s-r, 0) FA = max (r-s, 0) CR = min (1-s, 1-r)

Computation of Fuzzy Hit and False Alarm Rate

H= Σ(Hi)/ Σ(si) for i=1 to N

M= Σ(Mi)/ Σ(si) for i =1 to N

FA = Σ(FAi)/ Σ(1-si) for i=1to N CR = Σ(CRi)/ Σ(1-si) for i= 1 to N





As shown in Table 1, in a conventional SDT model, the Worker

would respond in a binary manner, either saying yes = 1, hazard or

signal present, and would stop work=0 or would respond no=0 (signal

absent) and carry on work=1.

Fig 4: Conventional SDT Table 1

Fig 5: FSDT Table 2

Both the signal and the response only take binary values. Even if the worker is uncertain, there is

always a binary response generated. Table 1, provides a truth table for conventional SDT, wherein

for all possible conditions the worker would yield a value that would populate only one out of the

four outcomes, and the rest would have zero membership. Fuzzy SDT, on the other hand, as

shown in Table 2, recognizes that the worker response is subject to an overlapping membership

in the two sets of ‘yes’ and ‘no’. Not all responses are distinctively clear between two states, just

like not all stimuli are distinctively clear between two states. There is a degree to which an event

is a signal, i.e., an unsafe condition, and a corresponding degree (for the same event) to which it

is a safe condition. Accordingly, there is a degree to which a signal present response (or, yes this

is an unsafe condition) is made, and a smaller degree to which the same response would be no,

signal is absent. Similar to conventional SDT, concepts of “Hit” (correct signal detection), “Miss”

(failed to detect a signal present), “False Alarm” (detected a signal when none present) and

“Correct Rejection” (decided no signal present when none present) are valid in the FSDT; it is

their binary characteristic that is discarded because of the loss of valuable information. Hence,

each event represented by a stimuli response pair in FSDT belongs, with some degree, to more

than one of the four categories used in conventional (or crisp) SDT. Consequently, it is possible

that events would claim nonzero membership in more than one outcome category, as shown in

Table 2(‘s’= degree to which an event is a signal; varies from 0 to 1; ‘r’= degree to which a ‘yes’

(signal present) response was made; varies from 0 to 1).

[5] Probabilistic neural network

The PNN is based upon Bayes' classification method.9 - 10' 11,12 The basis of the classification

method is given in Equation 1, where ht and hj are the prior probabilities, c,and Cj are the costs of misclassification, and ft(x) and f/x) are the true probability density functions:





hi Cffi (x ) hj , Cjfj (x). (1)

The difficulty with this relationship is that the prior probabilities (the probability that a sample

will come from a given population distribution) are generally unknown and must be estimated

from training data. This is done using Parzen's method of probability density function (PDF)

estimation. Bayes' classification will be more likely to group a new sample, x, into class / if the

prior probability or the cost of misclassification is high. This is especially important for

classifications where the cost of misclassification is not equal among the classes. In our

application, false alarms, the potential cost of misclassifying a nuisance is much more serious

than the cost for fire. In the PNN method different costs can be set for each class, thus producing

a better classification for those classes that demand higher performance. Finally, if the probability

density of a given class is large in the region of the new sample, x, then that class is favored. This

allows for multi-modal distributions to be dealt with appropriately when a nearest neighbor-based

classifier might fail. It has proven convenient and practical to implement the Parzen PDF

estimator in a neural network format, the PNN. The PNN is a nonlinear, nonparametric pattern

recognition algorithm that operates by defining a PDF for each data class based on the training

set data and the optimized kernel width parameter. For fire discrimination, the inputs are the

sensor responses or pattern vectors. The outputs of the PNN are the Bayesian posterior

probabilities (i.e., measures of confidence in the classification) that the input pattern vector is a

member of one of the possible output classes, for example, fire or nuisance source. The hidden

layer of the PNN is the core of the algorithm. During the training phase, the pattern vectors in the

training set are simply copied to the hidden layer of the PNN. Unlike other types of artificial

neural networks, the basic PNN only has a single adjustable parameter. This parameter, termed

sigma (a), or kernel width, along with the members of the training set, define the PDF for each

data class. In a PNN, each PDF is composed of Gaussian-shaped kernels of width CT located at

each pattern vector. The PDF essentially determines the boundaries for classification. The kernel

width is critical because it determines the amount of interpolation that occurs between adjacent

pattern vectors. As the kernel width approaches zero, the PNN essentially reduces to a nearest

neighbor classifier. A large kernel width has the advantage of producing a smooth PDF which

exhibits good interpolation properties for predicting new pattern vectors. Small kernel widths

reduce the amount of overlap between adjacent data classes. The optimized kernel width is a

compromise between an overly small or large a. Prediction of new targets using a PNN is more

complicated than the training step. Each member of the training set of pattern vectors (i.e., the

patterns stored in the hidden layer of the PNN and their respective classifications), and the

optimized kernel width are used during each prediction. As new pattern vectors are presented to

the PNN for classification, they are serially propagated through the hidden layer by computing

the Euclidean distance, d, between the new pattern and each pattern stored in the hidden layer.

The Euclidean distance scores are then processed through a nonlinear transfer function (the

Gaussian kernel) given in Equation 2:

Because each pattern in the hidden layer is used during each prediction, the execution speed of

the PNN is considerably slower than some other algorithms. The mass data storage requirements





can also be quite large since every pattern in the hidden layer is needed for prediction. Several

researchers have developed modified PNN algorithms to overcome this limitation, but were not deemed necessary for this application.

Fig 6 :Schematic of PNN network Designed in Matlab

PNN is an integral part of fusion between fuzzy logic and nueral network. There are 3 Layers in PNN

1) I/P:Where all the neurons are present

2) Hidden Layer: Processing of neurons for FAR detection

3) O/P:In this layer the FAR results are

[6] Hybrid Theory





Fig 7 :Flow chart of Hybrid Learning Theory

To reduce the false alarm system we are going to use the hybrid theory in which combines the two factors one is PNN and another one is Fuzzy logic.

1 .Sensor Status of Fuzzy logic

2. Apply status to fuzzy logic algorithm

3. FAR Output Generation

4. Sensor Status of PNN

5. Apply status to the PNN algorithm

6. FAR output Generation

7. FAR Output PN

8. Then control action

9. Else Correct rejection

[7] Implementation Details

Matlab:

MATLAB is a programming language developed by Math Works. It started out as a matrix

programming language where linear algebra programming was simple. It can be run both under

interactive sessions and as a batch job. This tutorial gives you aggressively a gentle introduction

of MATLAB programming language. It is designed to give students fluency in MATLAB

programming language. Problem-based MATLAB examples have been given in simple and easy

way to make your learning fast and effective.

Net beans:-





NetBeans is an open-source integrated development environment (IDE) for developing

with Java, PHP, C++, and other programming languages. NetBeans is also referred to as a platform of modular components used for developing Java desktop applications.

Figure 8:Real Time Data

The above image are the implementation images demonstrates our project in which real

time data give us the data from the sensor from the real time data

Figure 9 Matlab UI

The above image show us that the sensors connected to the Raspberry PI are modeled through

the Matlab UI by clicking on the sensor status we’ll be able to get the status of each and every

sensor by clicking on clear we‘ll be able clear the status and by clicking on fuzzy, we get a graph

as below





Figure 10: Crisp value

The graph above shows us the crisp value over here initially without applying the fuzzy logic

50% False alarm was present after applying fuzzy the slope tends to 20% i.e False Alarm Rate

has been decrease by 30%

Figure 11 :Crisp value

After Applying PNN - Fuzzy False Alarm Rate has been reduced by 50%

Figure 12:Raspberry PI

The Raspberry Pi 3 Model B is the latest version of the Raspberry Pi, a tiny credit card size

computer. Just add a keyboard, mouse, display, power supply, micro SD card with installed Linux

Distribution and you'll have a fully fledged computer that can run applications from word

processors and spreadsheets to games.we have also used sensors Moisture,Flame,smoke

[7] RESULTS





Figure 13:FAR Graph

The above graph consists us Total Attempt,True Positive,False Positive, False Negative

Where it is described as the total number of hits as the total attempts, true positive as hit and

found,True negative as Hit and Not Found, False positive as Miss and Found,False negative

as miss and not found

While applying base case,there are total number of 10 hits in which 5 true positive and 5 are

False negative which there are 5 hits and 5 miss as it states that there is 50% of False alarm is present in the current system

While applying the fuzzy algorithm, there are total number of 10 hits in which 5 are true

positive 3 are false positive and 2 are false negative i.e. after the application of the fuzzy algorithm as it states that the False alarm has been reduced by 30%

While applying the Fuzzy PNN algorithm, there are total number of 10 hits in which 5 are true

positive 0 are false positive and 5 are false negative i.e. after the application of the fuzzy-PNN

algorithm as it states that the False alarm has been reduced by 50% we have been able to enhance the false alarm rate by 50 %

Figure 14: Accuracy graph

The above graph states that the after applying and before applying we come to optimize the False

alarm reduction as there are 4 cases hit,miss,false alarm, correct rejection based upon which the

graph is plotted where it states that the base case gives us 50% accuracy ,After applying Fuzzy





accuracy level increases 20% i.e accuracy 70%,After Applying Fuzzy-PNN accuracy level

increases by 30%i.e Accuracy is 100%

[8] CONCLUSION

As the fuzzy logic has been aroud a while which has been proven efficient An overview of the

fuzzy logic based control are presented in this paper. The paper presents the basic principles of

the fuzzy logic and its difference to the conventional logic theory. The basic terms of fuzzy

logic, i.e. fuzzy sets, fuzzy set operations; fuzzy rules as also PNN contributes to the field of

technology through various factors PNN also helped us to detect fire via images and video as

our motive is to reduce the false alarm system by combining the to import aspect i.e PNN and

fuzzy logic Through which the false alarm reduction has been input has been taken from both

,after the unification of the both outputs the union sets has been applied to the output due to

which the false alarm reduction technology has been efficiently enhanced

[9] REFERENCES

[1] J. Tavares, F.J. Velez, J.M. Ferro, “Application of Wireless Sensor Networks to

Automobiles,” Measurement Science Review, vol. 8, Section 3, no. 3, 2008.

[2] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci, “Wireless sensor networks: a

survey,” Computer Networks (Elsevier) 38 pp. 393–422, 2002.

[3] Mirjana Maksimovi,etal “Developing a fuzzy logic based system for monitoring and early

detection of residential fire based on thermistor sensors” Computer Science and Information

Systems 12(1):63–89,2013

[4] E. H. Mamdani and B. R. Gaines, Fuzzy Reasoning and its Applications, New York:

North-Holland, 1985.

[5] A. Kaufmann and M. M. Gupta, Introduction to Fuzzy Arithmetic: Theory and

Applications, New York: Van Nostrand Reinhold Company Ltd., 1985.

[6] D. Dubois :Industrial Applications of Fuzzy Control, New York: NorthHolland, 1985.

[7] S. Daley and K. F. Gill, “Comparison of a Fuzzy Logic Controller with a P D Control

Law,” Transactions of the ASME Joumal of Dynamic Systems, Measurement, and Control, 111, June 1989.

[8] A. Ollero and A. J. Garcia-Cerezo, “Direct Digital Control, Auto-tuning and Supervision Using Fuzzy Logic,” Fuzzy Sets and Systems, 30, 1989.

[9] S. Tzafestas and N. Papanikolopoulos, “Incremental Fuzzy Expert PID Control,” IEEE

Trans. Ind. Electron.. 37, 5, October 1990.





[10] Japan Computer Quarterly, No. 79, Japan Information Processing Development Center, Tokyo, 1989.

[11] Cordón, O., Herrera, F., Hoffmann F., Magdalena L.: Genetic Fuzzy Systems: Evolutionary Tuning and Learning of Fuzzy Knowledge Bases. World Scientific Publishing:

Singapore, 19, (2001)

[12] Kapitanova, K., et al.: Using fuzzy logic for robust event detection in wireless sensor

networks. Ad Hoc Netw., (2011)

[13] Kumar Das, S., Kumar, A., Das, B., Burnwal, A.P., Ethics of Reducing Power Consumption in Wireless Sensor Networks using Soft Computing Techniques. International

Journal of Advanced Computer Research, 3(1-8), (2013)

[14] Reshma Shinde, Ritika Pardeshi, Archana Vishwakarma, Nayan Barhate “Need for Wireless Fire Detection Systems using IOT”(2017)

[15] Md Iftekharul Mobin , Md Abid-Ar-Raf , Md Neamul Islam , and Md Rifat Hasan “An

Intelligent Fire Detection and Mitigation System Safe from Fire” International Journal of Computer Applications (0975 - 8887) Volume 133-No.6,January 2016

[16] P. Singhala , D. N. Shah , B. Patel “Temperature Control using Fuzzy Logic” International Journal of Instrumentation and Control Systems (IJICS) Vol.4, No.1, January 2014

[17] Bhagyashri Nagorao Dhondge ,Suhas Sayajirao Jahdav “IOT Based Home Automation Using RASPBERRY PI” International Journal on Recent and Innovation Trends in

Computing and Communication ISSN: 2321-8169 Volume: 6 Issue:1

[18] Qassim, Qais; Patel, Ahmed; Mohd-Zin, Abdullah “Strategy to reduce false alarm in intrusion detection and prevention system” International Arab Journal of Information Technology

(IAJIT). Sep2014, Vol. 11 Issue 5, p500-506. 7p Prevention

[19] Susan L. Rose-Pehrsson , Ronald E. Shaffer a , Sean J. Hart “Multi-criteria fire detection

systems using a probabilistic neural network” a Naval Research Laboratory, Chemical Dynamics and Diagnostics 8 February 2000.


real time multisensor technique for false ......2018/06/02 · building automation infrastructure,...

Documents