Volume 3, Issue 7, July – 2018 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

An Explanation and Proposal on Multisensor Data

Fusion
Sathish Kumar S V
UG Student - Department of Mechatronics,
(Kumaraguru College of Technology, Coimbatore - 641049.)

Abstract:- Multi-sensor fusion (or) multi-sensor  Shorter response time.

information fusion is an emerging technology which is  Improved resolution.
being applied in the field of robotics, image and signal
processing and medical diagnosis. The main objective of Some of the key issues are
this paper is to give an idea about the various sensor  Nature of the sensors types available with resolution.
fusion performance and technical characteristic obtained  Ability of digitalization of sensors.
from different techniques. It is based on the principle of  Ability to implement the algorithm to the sensors and to the
integration of data from different sensors that could centre where it is controlled.
allow a better understanding of the data from different
varieties of sources for achieving better performance in II. CONCEPT OF MULTI SENSOR FUSION
many of the individual sources like weather forecasting,
statistical data analysis and estimation. In engineering The multisensor fusion concept is simple which has the
system, the fusion methods are significant because the basic four level process as follows.
system could provide capability to systems with different
Information Source: This is source of information about
sensors especially beyond that individual system of the
the sensors and other database which helps in situating the
sensors. Multi-sensor fusion of data allows integration of
sensors in the required place.
data from various sensors for improvising the perceiving
of environment and makes easier for decision making, Source Preprocessing: This stage helps in data pre-
planning, executing and control of automation. From screening and data allocation for the sensors which are going to
this paper had proposed an idea about the smart home be used. The helps the fusion process to checked before the
system with the help of various sensors which help to time.
mentor the home which helps the people for mentoring
the old persons in home. Level1 Processing (Object Refinement):
At this level the entities like position, velocity, identity is
Keywords:- Multi-sensor data fusion, image fusion, neural obtained which help for the military to target their enemies.
network image fusion, decision level fusion, integrating
This process involves four basic elements are data alignment
data, Smart Home.
(transformation of data to a consistent reference frame and
units), association (using correlation methods), tracking actual
I. INTRODUCTION and future positions of objects, and identification
Multi-sensor fusion is like an animal which evaluate Level2 Processing (Situation Refinement):
its surrounding with the help of the sounds (Signals) from At this level it helps to analyse some prior information
the surrounding which helps it to find the its surrounding is which should be mentored and want to immediately inform the
fit for it or not. The multisensor data fusion is like the user like objects, event and the context information’s.
human brain which helps us to recognise the sense of
different taste this is also like the animal sensing the sounds Level 3 Processing (Threat Refinement):
from the surrounding. Multi-sensor fusions also plays a Based on a prior knowledge from the level 2 and
major role in combining of sensor data from various sources predictions about the future situation helps in assessing the
for inferring the reliable and accurate results that is current situation of place. This is a quite difficult level of
unfeasible from the individual sensor. processing because it is not based on the figured database in
the system rather the things happen other than that like
The process of sensor fusion integrates data’s in such a strategies, environmental threats etc.
way that it results with the best performance which could be
attained if each information is used alone. From this Level 4 Processing (Process Refinement):
proposal it will give an idea about the smart home from the This process is a meta-process concerned with other
multisensor process in homes. The smart home is based on process. It helps in controlling the other system by monitoring
the motion detections and it helps in theft protection which the performance, its ability of working and identifying the
indicates the detection to the user. Some of the factors that information which is a feedback from the system which had to
improve the performance of the system. be improved. In this process it results in the objective of the
multisensory data fusion which we are targeting.
 Improved System reliability.
 Extended coverage.
 Improved confidence.

IJISRT18JL322 www.ijisrt.com 664

Volume 3, Issue 7, July – 2018 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Database Management:  Feature In – Decision Out:
This is act the brain for the process in which helps in Here the input is accepted as the feature format and the
storing all the data in the system. It helps in storing, retrieving, generates the desired decision output. Here the input is get
archiving, compressing, queries and protecting the data. This is from the sensors and the output which is generated as the
complex because we can’t predict the happening in the process, decision is displayed to the user. This type is mainly used in
so it is a complex process. many pattern recognition tasks like object recognition.

Human Machine Interaction:  Data In – Decision Out:

This process provides the interface between the system In this method the input is given as the raw data from the
and the human by the input and the communication to the sensors and the output is generated as the decision. This is like
fusion results to the operator and user. This interaction helps feature in – decision out mode.
the user to know about the information from the input and the
incident which taken place. Some model for the human and IV. HOME MONITORING SYSTEM
user interaction are JDL, Waterfall Fusion Process Model,
Boyd Model, The LAAS Architecture, The Omnibus Model. This is about the home security system, which helps the
user to continuously have an coverage of his/her home. It is
III. INPUT OUTPUT MODES also used as the monitoring system for the elder people using
the motion detectors like camera etc. Here some of the simple
If the input is given to the system, the output must be get sensors are used and the multi sensor fusion is implemented
from the system. So, the fusion process of input and output are using it’s information which makes the user to have the
done in six various modes. information about their home when they are out. Here we are
using some of the listed level 0sesnors:
 Data In - Data Out:
In this method the input and the output are in the form of  Open-Close sensor
data. This type of method is majorly used in the fusion process.  Motion detector
This method used in the front end of the processing stream and  CO detecting sensor
it works basically in the signals and image processing systems.  Glass breaking sensor
 Smoke detector
 Feature In – Feature Out:  Thermal Camera for detection
In this method it based on the features which is obtained
from both input and output. It occurs in the middle of the These sensors are referring to the level 0 sensor these
processing stream. Here the information which are got as a raw sensors are the basically used in the home monitoring system
measurement is then combined into a qualitative and which helps in a good monitoring guide for the user. The
quantitative. thermal sensors are used because it helps in easily monitoring
of the home also it helps the user to have a complete
 Decision In – Decision Out: monitoring about the health of the elders.
It is based on the decision for the input and the output. It
mainly occurs at the end of the processing stream. It is the In this level the sensors are positioned in the required
integrational process of the decisions from the different sensors position in the house which helps in observing the house by the
and data may be raw or extracted from the feature. This method user. Here my idea to use the thermal camera for the motion
is adopted in the instance when the used sensors are detection. These are basic sensors and idea for the simple home
compatible. security or monitoring system.

 Data In – Feature Out:

In this method the input is based on the raw data from the
sensors or other input facilities and the output is as the feature
format (visually representing it) with the surroundings or the
phenomenon we consider.

IJISRT18JL322 www.ijisrt.com 665

 Volume 3, Issue 7, July – 2018 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

O7 O6

BEDROOM1
M3

KITCHEN

O8

O5
BEDROOM 2

O9

BATHROOM

C1

O4

O10

LIVING ROOM
DINING ROOM ENTRY M2

M1
O3

O1 O2
O11

WINDOW

O1 GLASS SENSOR

M1 MOTION DETECTOR

C1 CO SENSOR

LEVEL 1

IJISRT18JL322 www.ijisrt.com 666

Volume 3, Issue 7, July – 2018 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
LEVEL 2: which gradually applies brake based on the distance of the
In this level the prior information about the location of obstacle before it but in traditional logic it’s like sudden
the sensors are loaded into the system containing the database. braking system when the obstacle is there. From this proposal a
In this level the complete details are provided about the perfect person can be able to understand the multi sensor data fusion
location of the sensors which make the user to understand the process with the proposal about the home monitoring system.
interfacing fusion process.
REFERENCE
Let’s consider this level from the proposal. We can see
the living room of the house it contains the motion detection [1]. Varshney, P K ‘Multi sensor fusion’, IET Digital
sensor (M2) and the glass break and the forward and backward Library Volume 9, Issue 6, December 1997, p. 245 –
motion sensors helps in window for opening and closing (O3 & 253.
O2) and the sensor for motion detection (M3) and the window [2]. Wilfried Elmenreich, ‘An Introduction to Sensor
sensors (O6 & O5) are in the Kitchen. Also, the motion is Fusion’, Research Report 47/2001.
detected it helps for the user to look for the illness of the old. [3]. https://www.researchgate.net/profile/Wilfried_Elmenrei
Thus, like this we are adding the context about the sensor’s ch/publication/267771481_An_Introduction_to_Sensor
absolute position in the home. _Fusion/links/55d2e45908ae0a3417222dd9/An-
Introduction-to-Sensor-Fusion.pdf.
LEVEL 3:
This level is completely based upon the level 1 and level [4]. Ren C. Luo and Chih-chenyih, ‘Multisensor Fusion and
2 information what we had considered which means the current Integration: Approaches, Application and Future
situation and the future prediction that makes the user to Research Directions’.
identify the issues in it. This level is completely based upon the
incidents which can be able to analyse the incidents which had [5]. Juan Go´mez-Romero, Miguel A. Serrano,
happened earlier and then now. [6]. Miguel A. Patricio, Jesu´ s Garcı´a, Jose´ M. Molina,
‘Context-based scene recognition from visual data in
Assuming the situation that the windows had been smart homes: an Information Fusion approach’.
broken. Let’s consider the window (O9) is first broke and then
the window (O6) after some months and then the window (O7) [7]. Jiang Dong, Dafang Zhuang, Yaohuan Huang, Jingying
after some days of the second window broke. This had been Fu,’Advances in Multi-Sensor Data Fusion: Algorithms
stored in database of the system which helps the in predicting and Applications’.
the future incidents when it may happen and the reason for the
breakage that helps the user for the clear solution about the
problem which he/she faces. Here in some instance the old age
people who are is also noted by their abnormal motion in the
home and this helps the user to call for emergency.

LEVEL 4:
This which act like the feedback loop system which
makes the user more comfortable for the incidents that happens
in the home which will help in giving the solution to the user
for the problem. From the diagram the dotted redlines marks
the privacy region of the house which indicated that it is the
higher risk region from the front door and the windows facing
the street.

From the assumption lets finalise that the sensors O11,

O1, O2 are at lower risk and it must have a slight importance
than the other. The other sensors from O3 to O10 are under the
privacy region which makes the user to have a continuous view
on and mainly the sensors which have been broke previously
O7, O6, O9 which are at higher risk which must be
continuously notes. In these cases, the fusion process helps in
for the better process and guide the user for the better updates.

V. CONCLUSION

Advancement in technology and software are the major

reason for the emerging of new technologies in the real life.
These technologies which doesn’t follow the traditional
concepts (1’s & 0’s) it is based on the fuzzy logic and the
neural network. The best example for the multi sensor fusion is
autonomous vehicles braking system it follows the fuzzy logic

IJISRT18JL322 www.ijisrt.com 667