See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323167491

Application of RFID and Path Planning Algorithm for Locating Products in a

Wholesale Market

Conference Paper · December 2017

DOI: 10.1109/ICAC3.2017.8318787

CITATION READS
1 160

4 authors, including:

Debajyoti Mukhopadhyay Trupti Baraskar

University of Mumbai Maharashtra Institute of Technology
220 PUBLICATIONS   852 CITATIONS    2 PUBLICATIONS   8 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Age Driven Automatic Speech Emotion Recognition System View project

Ontology based Decision Support System for Agriculture in India View project

All content following this page was uploaded by Debajyoti Mukhopadhyay on 14 February 2018.

The user has requested enhancement of the downloaded file.

Application of RFID and Path Planning Algorithm for
Locating Products in a Wholesale Market
Debajyoti Mukhopadhyay Trupti Barsakar Anuja Inamdar
Department of Information Technology Department of Information Technology Department of Information Technology
Maharashtra Institute of Technology Maharashtra Institute of Technology Maharashtra Institute of Technology
Pune, India Pune, India Pune, India
[email protected] [email protected] [email protected]

Vaidehi Yelnare Nirali Gundecha

Department of Information Technology Department of Information Technology
Maharashtra Institute of Technology, Maharashtra Institute of Technology,
Pune, India Pune, India
[email protected] [email protected]

Abstract—- Inventory management and supply chain are areas research is going on in RFID which was visible through varied
which have strong potential to benefit from Radio Frequency applications studied during the literature survey.
Identification technology. In the premises of a wholesale market, Optimum path planning has been an issue for concern in the
location tracking of a product itself becomes challenging for the fields of game development, vision, etc. The studied work
buyers. The implemented application will help the buyer to locate
revealed how traditional algorithms of path planning have been
required product and find the shortest path from his location to
the product’s location. This will help him to locate the product in improvised over these years to attain better performance and
optimum time, thus saving his efforts in searching the product efficiency. Designing of such a system which was application
manually. The location tracking is developed with the RFID specific involved considering real-time scenarios. The system
technology and shortest path is obtained through Digital Image was designed with the intention of making it easy to use for the
Processing. end user but at the same time fulfilled the requirements.
Conventionally barcode system was widely used, but
Keywords—Radio Frequency Identification (RFID), A* considering its limitations RFID evolved to overcome its
algorithm, Location Identification, Internet of Things (IoT), shortcomings. RFID extends capacities of barcode system to
Optimum Path finding, Message Passing Interface (MPI), Parallel read data dynamically. The proposed system will help the buyer
Bidirectional Search (PBS), Received Signal Strength Indicator
(RSSI), Graphical User Interface (GUI), Radio Frequency (RF),
to locate his required product and the shortest path from his
Attention (AT), Electronic Product Code (EPC), System-On-Chip location to the product’s location in the premises of a wholesale
(SOC). market. This will help him locate the product in less time, thus
saving his e orts and time in searching the product manually.
The location tracking will be realized with the RFID technology
I. INTRODUCTION and optimum path will be obtained. The system will be divided
With the advancements in networking a new field of Internet of into three parts. First would be location tracking of user and the
Things (IoT) was born. Human kind since ages are working to product. This will be done by attaching RFID tags to both and
make mechanical tasks automated. The requirement of the reader will provide the estimate location. In the latter part,
automation in smart devices which are capable of sensing data the location of source and destination will be mapped onto the
environment and processing this information and image of the shop layout. Optimum path will be found using A*
communicating it with other devices. IOT with the backbones algorithm using heuristic functions. The popularity of products
of Wireless Sensor Networks (WSN) and RFID provides a can also be measured with the help of product search logs.
framework to build automated and smart systems. Therefore, the system is a RFID based navigation system in
Applications of IOT for automation are varied and increasing wholesale shop.
at a high pace with the innovations of human minds. One such The following paper is ordered as follows: Section II
scenario is searching products in a wholesale market. In a huge represents a short-lived review of the related literature. Section
storage of diverse products, locating and finding a way out to III which discusses the proposed system for locating and
reach them becomes a tedious and time-consuming task. The finding an optimum path in wholesale market. The
system proposed and implemented here aims to solve this experimental results are presented in Section IV and section V
problem. The implemented system uses RFID for localization covers the testing and result analysis. Finally, the conclusions
and identification of products and user. In recent years, lots of are whole up in Section VI.

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

 II. LITERATURE SURVEY tags, mapping them onto the shop layout image, computing
The first part of literature survey is Path Planning and the the shortest path and giving that as the result to user through
second Part of Literature Survey is RFID Based Indoor GUI. Server layer is responsible for sending the list of
Positioning System. MPI implementation based on cluster available products to user, localization of the user and
systems that identify the closest vertices to the source vertex product tags, obtaining the data about tags from hardware and
case study was done in University of Buffalo. The conclusion storing it and maintaining the database. It acts as an interface
from this study was that Dijkstra’s algorithm is better for between hardware and GUI. The user application and server
execution time in each case. Although there are better communicates using Wi-Fi. The hardware layer consists of
solutions to find out the shortest path, Dijkstra’s Algorithm is the actual RFID readers which are controlled by Arduino
used in systems that require stability [1]. Time complexity of Uno. The RFID readers reads the nearby RFID tags. The
Dijkstra’s is O(n2) which isn’t of significance when the Arduino with the help of Wi-Fi sends this data to server for
source is moving, actual operation speed needs to be taken localization. Hardware, server and user application
into consideration. The concept of priority region was communicates using Wi-Fi.
implemented to overcome the problem of computation speed The basic functionality of system is achieved by the three
[2]. A* algorithm is better than other path finding algorithms building blocks of system, each playing its own vital role as
as it doesn’t traverse the entire nodes of the maps. Improvised follows.
version of A* algorithm includes weighted processing and
introduction of manual searching marker to avoid falling into
the dead state, which results in less iterations, search cost and
process time [3]. The Parallel Bidirectional Search (PBS) is
an alternative to optimized shortest path algorithm. PBS A* Fig. 2. Hardware components
was implemented by running 2 threads from start node and
finish node. This test reflected that PBS A* consumed less A. Hardware:
time for execution (68.8% faster than not optimized A*) [4]. The hardware blocks implementation reflects applications
The studied positioning Kalman-Filter was developed to of RFID. RF communication occurs by the transference of
estimate RFID reader position based on instantaneous RSSI data over electromagnetic waves. By generating a specific
measurements received from tag of neighboring readers. This electromagnetic wave at the source, its effect can be noticed
positioning system can be implemented for navigational at the receiver far from the source, which then identifies it
applications as the postulation calls for instantaneous RSSI and thus sends back the information. The hardware block is
measurements [5]. divided into four modules as follows:

III. PROPOSED WORK 1) RFID Reader Module: The RFID reader module
consists of an antenna, transceiver and decoder, thanks...”.
The basic functionality of the proposed system is Put sponsor acknowledgments in the unnumbered footnote on
checking the availability of the products, finding their the first page which sends periodic signals to inquire about
location and providing the user with an optimum path
any tag in vicinity. On receiving any signal from a tag it
between user and the required product. The proposed system
passes on that information to the Arduino Board. The
architecture takes into the consideration the requirements for
MFRC522 RFID Reader module has been used which is
providing the functionalities.
comparatively cost efficient and readily available in the
market.
2) Passive RFID Tags: Tags can be either active or
passive. The passive tags use the power induced by the
magnetic field of the RFID reader. The RFID tag is composed
of an antenna, a wireless transducer and an encapsulating
material. The tags used are rewritable and so coded with an
electronic product code (EPC) which holds the information
about product. Apart from EPC, every tag comes with an
unique identifier. These tags are read by RFID reader. Based
on EPC it is decided whether it is user card or product tag.
Product tags are attached on products and customers are
given RFID cards. Wi-Fi Module: The Wi-Fi module
Fig. 1. System Architecture ESP8266-01 has been used for sending the data to server. The
ESP8266 Wi-Fi Module is a self-contained SOC with
The system has a three-layered architecture. As shown in
integrated TCP/IP protocol stack that can give any
fig.1, user application layer is responsible for taking inputs
microcontroller access to Wi-Fi network. It has a pre-
from user, sending it to the server, obtaining the locations of
programmed firmware which enables connecting Arduino different locations inside the lab. The image of layout of the lab
Uno board to Wi-Fi using AT commands. was created based on the arrangement of components. All five
hardware modules, server and Android phone were connected to
3) Arduino Uno board: The Arduino Uno Board has been
a Wi-Fi router. The product tags are attached on products which
programmed with the help of Software Serial library which
are kept near RFID readers. The customer is handed a RFID
enables serial communication on other digital pins of the
card. Hardware modules were placed at the location A, B, C, D
Arduino board apart from the two pins which are initially
and E.
assigned. The RFID reader reads the RFID tags and sends this
data to Arduino and Arduino forwards this data to Wi-Fi
module. The Wi-Fi module communicates with the server
using POST method.

B. Server:
The server holds database. All the hardware modules and
user applications communicates with server using POST
method. MySQL along with PHP has been used to accomplish
Fig. 3. Experimental Setup
server goals. The functionality of server can be divided into three
parts based on who they communicate with. 1) Wi-Fi Router: The Wi-Fi router used is of brand D-Link
having model number DIR-61.
1) Server and User Application Communication: The server
is responsible for authenticating user’s card number, providing 2)Server: Laptop which holds server is Lenovo S20-30
the list of available products and providing the location of user which has Intel Celeron N2840 @ 2.16 GHz Dual Core
and product. Processor, 2GB RAM, Wi-Fi enabled and operating system is
Ubuntu 16.04 LTS.
2)Server and Hardware Communication: The server is
responsible for receiving the data sent by hardware unit, 3)Android Mobile: Android Mobile used is Sony Xperia
identifying whether it is a product tags information or user tags C6802which has 2.2GHz quad core processor coupled with 2GB
based on which it stores the data in appropriate tables. of RAM and Wi-Fi enabled.
3) Server and Server’s Graphical User Interface: The 4) RFID Reader: The MFRC522 is a highly-integrated
management team of the wholesale market has access over the reader/writer IC for contactless communication at 13.56 MHz
server’s front end. The team includes Head and Manager. The the MFRC522 reader supports ISO/IEC 14443 A/MIFARE and
front end provides a platform for the management team for task NTAG.
such as view inventory, view the analysis based on search logs
5) RFID Tags: 13.56 MHz RFID tags which are rewritable.
and purchase logs. The individual from management team has
authorities over the data based on their designations. The head is 6)Wi-Fi Module for hardware is ESP8266-01 which has
head of all the managers. He approves the requests of managers. Integrated TCP/IP protocol stack, Wi-Fi 2.4 GHz, support
Head views the details of the wholesale market. The manager WPA/WPA2.
can view all the details of the wholesale market and create his
account. V. TESTING AND RESULT ANALYSIS
The system was tested using two algorithms for optimum
C. User Application:
path finding, Dijkstra’s and A-star. Dijkstra’s algorithm works
The user application has been developed using Android. by visiting vertices in the graph starting with the object’s starting
Emphasizing on the implementation of functionality along with point. It then repeatedly examines the closest not-yet-visited
taking utmost care of making it user friendly, it comprises of vertex, adding its vertices to the set of vertices to visit. It expands
three activities. The application is lightweight and has been outwards from the starting point until it reaches the goal.
designed to work on minimum SDK version 15 and target SDK Dijkstra’s algorithm is guaranteed to find a shortest path from
version 24. User enters his card number and sends it to the server the starting point to the goal, as long as none of the edges have
for authentication. On successful authentication, the application a negative cost. A-star algorithm which is a greedy best first
fetches the list of available products and displays it on screen. search algorithm works in a similar way, except that it has some
On selection of product, the user is provided with an optimum estimate (called a heuristic) of how far from the goal any vertex
path to reach the product. A-star algorithm is used for optimum is. Instead of selecting the vertex closest to the starting point, it
path finding. selects the vertex closest to the goal. Greedy best first search is
not guaranteed to find a shortest path. However, it runs much
IV. EXPERIMENTAL SETUP quicker than Dijkstra’s algorithm because it uses the heuristic
The experiment was conducted in Research Laboratory of function to guide its way towards the goal very quickly. So here
MIT College, Pune. Hardware units were physically placed at the tradeoff is between time and accuracy. The heuristic function
h(n) tells A* an estimate of the minimum cost from any vertex n parameters such as time required, cost of path that was found,
to the goal. If h(n) is 0, then only cost g(n) plays a role, and A* iterations and time it took.
turns into Dijkstra’s algorithm, which is guaranteed to find a
shortest path. If h(n) is always lower than (or equal to) the cost
of moving from n to the goal, then A* is guaranteed to find a
shortest path. The lower h(n) is, the more node A* expands,
making it slower. If h(n) is exactly equal to the cost of moving
from n to the goal, then A* will only follow the best path and
never expand anything else, making it very fast. Various
Heuristic functions have been proposed based on the
representation of graph and direction of movements that are
allowed. A bitmap image of layout of wholesale market is first
Fig 4. Path Found from A – C Fig 5. Path found from A-C
gridded into a cell of 30*30 pixels each. The image is now using Dijkstra’s Algorithm using Dijkstra’s Algorithm
converted into a grid of 60*60 cells, each cell is a node in a (Grid View)
graph. The cost of adjacent movements is considered 1 unit and
diagonal movement as 1.4142 units. Heuristic functions used are
Euclidean distance and Octile distance.
1) Euclidean Distance: The straight-line distance between
source and destination is given by

=
( − ) ( − )
(1)
Fig 6. Path Found from A – C Fig 7. Path Found from A-C using
ℎ , , using A-Star algorithm with A-Star algorithm with Euclidian
Euclidean Distance Distance (Grid View)
,

2) Diagonal Distance: This is applicable for graphs whose

nodes exhibit 8-point connectivity.

= ×( ) ( −2 × )
× min( , )
(2)
ℎ ,
= ℎ
= ℎ Fig 8. Path Found from A-C Fig 9. Path Found from A-C
= ( − ) using A-Star algorithm with using A-Star algorithm with
= ( − ) Octile Distance Octile Distance (Grid View)
,
,

ℎ =1 = √2 ,
.

3) Results and Analysis for different images:

The working of A* was analyzed based on cost of path it finds,
iterations required and time it took to complete the search. A
bitmap image of layout of wholesale market is first gridded into
a grid of 30 30 pixels. The image is now converted into a graph
Fig 10. Bitmap layout 1 Fig 11. Bitmap layout 2
of 50 60 grids, each grid is a node in a graph having 8-point
connectivity. We considered 2 different layouts and observed
 TABLE I. OBSERVATIONS FOR LAYOUT 1

Heuristic Function = 0 Heuristic Function = Heuristic Function =

Sr.
Source Destination (Dijkstr’s Algorithm) Euclidean Distance Octile Distance
No.
Cost Iteration Time (ms) Cost Iteration Time (ms) Cost Iteration Time (ms)
1. B A 42.79 1848 126 42.79 1160 92 42.79 1134 82
2. A B 42.79 1634 111 44.45 1124 96 44.45 1110 81
3. A C 28.24 994 75 28.24 746 72 28.24 716 54
4. C A 28.24 1219 92 28.24 566 60 28.24 537 56
5. B C 26.65 1023 82 26.65 659 63 26.65 620 50
6. C B 26.65 1095 84 26.65 552 53 26.65 552 79

TABLE II. OBSERVATIONS FOR LAYOUT 2

Sr. Source Destination Heuristic Function = 0 (Dijkstra’s Heuristic Function Heuristic Function
No. Algorithm) = Euclidean Distance = Octile Distance
Cost Iteration Time (ms) Cost Iteration Time (ms) Cost Iteration Time (ms)
1. A B 45.94 961 59 45.94 542 34 45.94 1134 82
2. B A 45.94 883 60 45.94 588 54 45.94 582 42
3. A C 18.31 246 25 18.31 178 20 18.31 175 17
4. C A 18.31 388 42 18.31 162 24 18.31 160 20
5. B C 34.21 968 59 34.21 485 54 34.21 412 38
6. C B 34.21 980 71 35.04 495 38 35.04 487 35
7. A D 44.69 934 69 44.69 614 43 44.69 600 35
8. D A 44.69 946 58 46.35 696 51 46.35 698 50
9. D B 21 486 33 21 304 41 21 288 32
10. B D 21 301 21 21 254 22 21 247 18
11. C D 30.48 907 7 30.48 545 48 30.48 525 46
12. D C 30.48 671 47 30.48 529 43 30.48 517 37
13. A E 35.04 628 63 35.04 381 34 35.04 378 28
14. E A 35.87 887 54 35.87 545 40 35.87 542 36
15. B E 37.52 724 43 37.52 446 30 37.52 446 30
16. E B 37.52 923 57 37.52 530 40 37.52 523 38
17. C E 20.24 488 45 20.24 24 35 20.24 23 25
18. E C 20.24 452 32 21.07 239 31 21.07 238 27
19. D E 25.89 584 39 25.89 384 31 25.89 368 28
20. E D 25.89 687 44 25.89 382 33 25.89 365 31
21. E D 25.89 687 44 25.89 382 33 25.89 365 31

In the above table I, locations are A(43,11), B(20,39) and C(16,14) and in table II, locations are A(7,11), B(43,11), C(15,26),
D(43,32) and E(20,39).

Analysis based on results of layout 1 and layout 2:

Fig 12: Analysis of testing results of image 1 Fig 13: Analysis of testing results of image 1
From the graphs and observed values for finding the path iterations which are much lower than Dijkstra’s, their
between two points, all three functions result in paths which may computation time is also less. Even though there is less
appear different visually but have same cost in almost all cases. difference between Octile and Euclidean distance, Octile
The performance of these functions is different because the Distance finds path in comparatively less iterations. A* doesn’t
number of iterations it takes for each one of them to find path.
always guarantee finding the optimum path but the number of
Dijkstra’s clearly takes higher number of iterations which results
cases where it fails to find optimum path are very less with the
in consuming more system time for computations, while Octile
distance and Euclidean distance take nearly same amount of right choice of heuristic function. A grid of 52*48 cells was
considered for an image of 1560*1440 pixels with no obstacles. range readers which will improvise the system to be
Observations were noted down for the case where source = implemented on a larger scale.
(0,0) and destination = (52,48) as follows.
VII. CONCLUSION
TABLE III. OBSERVATIONS FOR IMAGE WITH NO OBSTACLES
Based on the demand of wholesale market the presented
Heuristic Function Cost of the Path Found Iterations Time (ms) system is helpful for finding the location of products and an
0 71.88 2648 185 optimized path to reach that product. The product and user
localization is done with the help of RFID technology. Path
Euclidean Distance 71.88 1722 129 planning was done using A* algorithm with Octile distance as
Octile Distance 71.88 1701 119 the heuristic function. Results of testing clearly reveal the
success of the implemented system. Dijkstra’s algorithm, A*
algorithm with Euclidean distance and Octile distance as the
As we can see, out of the total 27 cases, only in two cases, A* heuristic function were implemented for path finding and tested
algorithm using Octile distance fails to find the optimum path. for various cases. Analysis of the results concluded that the
Accuracy of the algorithm is measured as stated below. accuracy of A* algorithm with Octile distance as the heuristic
function to find the shortest path was 92.60%, it was faster than
Dijkstra’s algorithm by 35.68% and iterations required were
(ℎ = ) = × 100 (3) much less compared to Dijkstra’s algorithm. For two cases the
ℎ , = paths found by A* algorithm was not shortest but were 96%
accurate.
= ℎ ℎ
∗
REFERENCES
ℎ =
27 − 2 [1] Bogdan Popa and Dan Popescu, “Analysis of algorithms for shortest
(ℎ = ) = × 100 path problem in parallel”, IEEE, Carpathian Control Conference
27 (ICCC), 2016 17th International on 29 May-1 June 2016, pages
= 92.60% 613-617, DOI:1109/CarpathianCC.2016 .7501169, ISBN: 978-1-
4673-8606-7
The A* algorithm with Octile Distance was successful in [2] Yin Chao and Wang Hongxia, “Developed Dijkstra Shortest Path
Search Algorithm and Simulation”, Computer Design and
finding optimum path accurately for 92.60 % of experimented Applications (ICCDA), 2010 International Conference on 25-27
cases. Also, the time required to find the longest path in a grid June 2010, Volume 1, pages VI-116 to VI-119, DOI:
with no obstacle can be considered to compare the algorithms. 10.1109/ICCDA.2010.5541129, ISBN: 978-1-4244-7164-5, (IEEE
Percentage of time difference was computed as stated below. Explore)
[3] Junfeng Yao, Chao Lin, Xiaobiao Xie, Andy JuAn Wang, Chih-
Cheng Hung, “Path Planning for Virtual Human Motion Using
( ) ( )
%∆ = × 100 (4) Improved A* Algorithm”, 2010 Seventh International Conference
( ) on Information Technology, 12-14 April 2010, Las Vegas, NV,
185 − 119 USA, ISBN: 978-0-7695-3984-3/10, DOI: 10.1109/ITNG. 2010.53,
= × 100 = 35.68% (IEEE Explore)
185
[4] Pratyaksa Ocsa Nugraha Saian, Suyoto, Pranowo, “Optimized A-
Star Algorithm in Hexagon-Based Environment using Parallel
A* algorithm using Octile distance was 35.68% faster than Bidirectional Search”, 2016 8th International Conference on
Dijkstra’s algorithm for this case. A* algorithm with Octile Information Technology and Electrical Engineering (ICITEE),
distance as the heuristic functions has proven to be efficient and Yogyakarta, Indonesia, ISBN: 978-1-5090-4139-8/16, DOI:
10.1109/ICITEED.2016.7863246, (IEEE Explore)
hence suitable for this application.
[5] Samer S. Saab, Senior Member, IEEE, and Zahi S. Nakad, Member,
“A Standalone RFID Indoor Positioning System Using Passive
VI. FUTURE WORK Tags”, IEEE Journals & Magazines, IEEE transactions on industrial
The implemented system is built using RFID readers which electronics, VOL. 58, NO. 5, MAY 2011, Pages: 1961 1970,
a have short read range, operating at the frequency 13.56 MHz DOI:10.1109/TIE.2010.2055774, (IEEE Explore)
and the RFID tags are passive. Use of active RFID tags could [6] Mathieu Bouet and Marie Curie Paris, RFID Tags: Positioning
possibly increase the read range but it will increase the cost of Principles and Localization Techniques, Wireless Days, 2008.
WD08.1st IFIF, Pages: 1-5, DOI:10.1109/WD.2008.4812905,
such systems when implemented on large scale. Powerful RFID (IEEE Explore)
readers which can detect RFID tags from long distance are [7] Ms. Ashwini W. Nagpurkar, Mr. Siddhant K. Jaiswal, "An
suitable for the application. Such readers can cover large area Overview of WSN and RFID Network Integration",2015, IEEE
and read multiple tags stored at different locations. The Sponsored Second International Conference on Electronics and
locations of such tags can be obtained using techniques which Communication Systems (ICECS '2015), DOI:
are based on the principles of distance estimation, scene 10.1109/ECS.2015.7124956, ISBN: 978-1-4788-7225-8/15/,
analysis and proximity. Paraments such as Time of Arrival, (IEEE Explore)
Received Signal Strength, Time Difference of Arrival,
Received Signal Phase, etc. are used in the algorithms for
localization [6]. Such methods can be implemented using long

View publication stats