DESIGN A REVERSE LOGISTICS INFORMATION SYSTEM

WITH RFID

Carman Ka Man Lee

, Tan Wil Sern

and Eng Wah Lee

School of Mechanical & Aerospace Engineering, Nanyang Technological University, Singapore

Singapore Institute of Manufacturing Technology, Singapore

Keywords: Reverse Logistics, Radio Frequency Identification (RFID), Genetic Algorithm, Location determination.

Abstract: Recently, reverse logistics management has become an integral part of the business cycle. This is mainly

due to the need to be environmental friendly and urgent need to reuse scarce resources. Traditionally,

reverse logistics activities have been a cost center for most businesses without generating extra revenue.

However, due to recent increase in commodity and energy prices, reverse logistics management could

eventually be a cost savings method. In this research, we propose using Radio Frequency Identification

(RFID) technology to better optimize and streamline reverse logistics operations. Using RFID, we try to

eliminate parts of the unknowns in reverse logistics flow that made reverse logistics model complicated.

Furthermore, Genetic algorithm is used to optimize the place of initial collection center so as to cover the

largest population possible in order to reduce logistics cost and provide convenience to end users. This study

is based largely on literature review of past workings and also experiments are conducted on RFID

hardware to test for its suitability. The significance of this paper is to adopt ubiquitous RFID technology and

Genetic Algorithms for reverse logistics so as to obtain an economic reverse logistics network.

1 INTRODUCTION

Reverse Logistics is the management of the return

flow of materials from end users back to the

producers. The management for reverse logistics is

the opposite of conventional supply chain flow.

Recently, reverse logistics management came into

focus in an effort to cut cost. As of 1999, total value

of merchandise returned in the U.S amounted to $62

billion. This represents a loss of around $10 -$15

billion to retailers while the cost of handling

returned products was estimated to be $40 billion

(ReturnBuy, 2000). However, the management of

reverse logistics is complicated due to the many

unknowns in the system and integration of reverse

logistics to the forward logistics also proves to be a

challenge (Fleischmann et al., 1997). This paper

aims to overcome the uncertainties in reverse

logistics process using wireless technology such as

Radio Frequency Identification (RFID) to enable the

dissemination of product information in real time.

Genetic Algorithm (GA) is widely used in various

aspects in supply chain to find out the optimized

solution (Min et al., 2003). In this paper, GA is

employed to determine the optimum location to

deploy the various initial collection centers to

develop an optimum reverse logistics network

linking initial collection point and centralized return

centers.

2 RELEVANT LITERATURE

2.1 Reverse Logistics

Reverse logistics is a field of interest with many

recent studies conducted and models developed.

Many models have been developed with regards to

reverse logistics management with Barros et al.

presenting a network for recycling of sands

(Fleischmann et al., 1997). In his model, he

proposed a multi-level capacitated warehouse model

while using scenario analysis to solve for the

uncertainties in the return flow. Spengler et al. on

the other hand developed a mixed-integer linear

programming model for recycling of industrial waste

(Fleischmann et al., 1997) based on a multi-level

capacitated warehouse location. At present, not

many studies have been done to formulate a model

integrating forward and reverse logistics process.

Most of the models currently are planned based on

forward logistics purposes. Min et al. (Min et al.,

2003) developed a decision support system to

293

Lee C., Sern T. and Lee E. (2009).

DESIGN A REVERSE LOGISTICS INFORMATION SYSTEM WITH RFID.

In Proceedings of the 11th International Conference on Enterprise Information Systems - Artiﬁcial Intelligence and Decision Support Systems, pages

293-299

DOI: 10.5220/0001856802930299

 SciTePress

support the design of distribution and collection

networks. By using the location of the facility, the

system can determine the optimal good flows in the

return network and the resulting costs (Fleischmann

et al., 1997). From these literature reviews, it is

shown that reverse logistics process is not

necessarily the opposite of forward logistics.

Reverse logistics is more complicated in a sense that

there are different actors involved and the

uncertainties in the system are higher. This has not

been addressed by most of the models at present and

hence we will try to solve this in this research.

2.2 Radio Frequency Identification

(RFID)

Radio Frequency identification (RFID) has been

invented long ago but only recently has the

technology emerge in mainstream applications.

RFID is mainly used to tag products for

identification purposes much like bar-code system

today. However, RFID offers value-added advantage

in that RFID does not need line-of-sight for

operation and can read multiple tags at the same

time, which has been a constraint for bar-code

technology (Want, 2007). Much research is being

done at the moment to search for application usage

of RFID with recent adopters being US retail giant

Wal-mart and Tesco. Although RFID promises a lot

of advantages, there are also constraints and

concerns about the technology. Most importantly is

the additional cost needed to implement RFID

technology to the current system because at the time

being, RFID hardware is still costly. Secondarily,

there is also a concern on the reliability of RFID

technology and finally concern for privacy (Hunt,

2007). At the time being, not many studies have

been done to integrate RFID technology to reverse

logistics management system and this is an area

where we will try to explore.

2.3 Genetic Algorithm

Genetic Algorithm (GA) is a form of mathematical

optimization technique. GA can be used in various

applications such as determining the optimum flow

for a factory production process, optimum traveling

route to determination of location of warehouses.

As compared to traditional optimization

techniques, genetic algorithm is more robust due to

the following features:

• Genetic algorithm works with a set of

parameter and not the parameter itself. This

lends to the robustness of genetic algorithm.

• Genetic algorithm search from a solution space

and not a single point. Hence, genetic

algorithm method does not depend on the

existence of derivatives like traditional

optimization techniques do.

• Genetic algorithm uses an objective function to

determine the score of a chromosome and not

based on derivatives or other auxiliary

knowledge.

• Genetic algorithm uses probabilistic transition

rules. This means that genetic algorithm selects

strings to be included in the next stage or

process based on probability whereby

chromosomes with higher score are assigned a

higher probability to be selected. This is the

basis of survival of the fittest law.

All these advantages of genetic algorithm make it

more robust and useful as compared to other

traditional methods.

3 PROBLEM DEFINITION

As stated earlier, reverse logistics management is

complex and involves a huge amount of unknowns.

In-depth study has been done in this field and

mathematical models have been developed to

enhance the efficiency of reverse logistics. However,

there are still limitations due to the following

reasons:

1. Reverse distribution network is full of

uncertainties such as the quantity and quality of

the returned products from end users.

2. Different actors involved in the reverse

distribution channel require different inventory

control mechanism.

3. Forms of reuse differ for each product

requiring different planning and coordination.

4. Difficulties in inventory control in systems

with return flow.

Furthermore, there are also problems associated with

integrating reverse flow of material into forward

logistics due to:

1. Reverse distribution is not necessarily

symmetric picture of forward distribution.

2. Special characteristics of reverse logistics

include:

• Many-to-few network structure

• System uncertainties

3. Returned products may be sent to original

producer or to third party recycling centre.

Hence, further uncertainties and unknowns are

involved.

ICEIS 2009 - International Conference on Enterprise Information Systems

294

4. Modification and extension may be necessary

to tailor for reverse distribution.

For illustrative purposes, let us suppose that a

computer manufacturer (called Apple hereafter)

selling hardware to end-users intends to recycle used

computers from its customers. Customers who wish

to return the computer will dispose their computer at

the various initial collection points and Apple will

employ logistics providers to collect the disposed

computer from the initial collection points.

Given the limited space at the initial collection

points, returned products should be collected and

shipped to the centralized collection centre as soon

as possible and the components collected should be

determined as soon as the returned products reaches

the centralized collection center.

At the centralized collection center, components

that can be reused may be redistributed after minor

cleaning or refurbishing while defective components

will undergo recycling processes. With the above

situation as a model, we wish to address the

following issues:

1. Location of initial collection point such as to

cover the largest possible population in an area.

2. Location of centralized collection center such

that transportation from initial collection point

to the centralized collection center is minimum.

3. Method to obtain component information most

efficient and timely from the returned product

to determine for its quantity and quality.

4 MODEL DESIGN

4.1 Framework of Reverse Logistics

Reverse logistics is the management of material flow

opposite to the conventional supply chain flow

(Fleischmann et al., 1997). It encompasses logistic

activities to transform used products from end users

back to usable products again.

Reverse logistics management involves 3 main

stages:

1. Distribution planning aspects – This stage

involves physical transportation of used

products from end users back to the producers.

2. Transformation – The recovery of returned

product back to usable product. There are

several ways of transformation including:

Direct reuse, Repair, Recycling and

Remanufacturing.

3. Inventory Management – This is to manage

inventory level and to integrate supply flow

from both the traditional supply chain and also

from reverse logistics (Fleischmann et al.,

1997).

Dimensions of Reverse Logistics. There are

many instances of reuse criteria. These can be

classified as motivational, items recovered, forms of

reuse and the actors that are involved in the process.

Reuse motivation can be due to economical and also

ecological. In terms of economical, usually

machinery parts can be reused with slight repairs

and this saves cost as compared to manufacturing a

new part. As for ecological concerns, companies are

increasingly being pressured to take back all their

sales materials for recycling purposes in order to be

eco-friendly.

As for different types of materials recovered, the

forms of reuse may vary as well. The different forms

of reuse includes:

• Direct Reuse. Returned materials can be

reused directly without major repairs except for

cleaning. Examples of such products are bottles

and containers.

• Repair. This process is to restore failed

products into working order. However, the

performance of the repaired product might be

reduced.

• Recycling. This process recovers material

without conserving any of the initial product

structure. Commonly recycled items are scraps,

paper and glass.

• Remanufacturing. This process differs from

recycling in that the recovered product retains

its original characteristics. Examples are

automotive engines and machines.

Finally, there are also different actors involved in

reverse logistics. Actors play different parts in the

reverse logistics process such as collection, testing

and product recovery. Due to the different actors

involved in reverse logistics, integrating reverse and

forward logistics pose a major challenge.

As stated in the problem definition section, there

are many challenges in designing an optimum

reverse logistics management system and integrating

forward and reverse logistics. Most of the problems

arose mainly due to uncertainties in the reverse flow

of materials both in terms of quantity and quality

and also the timeliness of the information gathered.

Below, we propose a model that integrates RFID

technology into the reverse logistics framework to

eliminate the uncertainties involved in the process.

Furthermore, we will also employ genetic algorithm

optimization technique to determine the optimum

location for the initial collection point so as to

maximize user coverage and reduce logistics cost

involved in the process.

DESIGN A REVERSE LOGISTICS INFORMATION SYSTEM WITH RFID

295

4.2 The Reverse Logistics Model

The physical flow and RFID data flow for reverse

logistics process are shown in Figure 1. The main

elements in reverse logistics includes end user,

intial collection points, centralzied collection center

and produccer.

Figure 1: Flow chart of reverse logistics process.

4.2.1 End Users

End users are customers of the company who wishes

to dispose off of their old products or are returning

their products due to defects. End users would

demand for convenience and accessibility. Hence,

the initial collection points have to be loacated

within short travelling distance to the customers.

4.2.2 Initial Collection Points

In order to determine the placement of the initial

collection points, we shall employ the genetic

algorithm optimization technique. The locations and

density of population in an area are recorded and

amount of initial collection center is determined.

Using the genetic algorithm program, the optimum

location to employ the initial collection center is

determined.

Genetic Algorithm. Genetic algorithm is one of

many population-based optimization techniques.

Population-based technique induces a pattern whose

present state depends on the past states; hence

incorporate implicit memory structure (Min et al.,

2003). Genetic algorithm is based on the theory of

evolution whereby an initial population is generated,

subsequently the population space are evaluated

against a pre-defined function to remove inferior

solutions. After that, mutation of strings will then

occur and this process repeats until the optimum

solution is found.

The Working Mechanism of Genetic Algorithm.

Genetic algorithm technique starts with

initialization. The close (0) or open (1) of collection

point(cp) is encoded in the chromosome For

instance:

… …cp

1 0 1 1 1 0 0 0 1

After the initialization phase, the chromosomes in

the solution space will be evaluated with reference

to a pre-defined objective function. The

chromosome with a higher score will be chosen to

form an intermediate solution space. This is called

the selection process.

For our reverse logistics determination of collection

point, an example of the objective function would be

as below:

Minimize

)XXX(

iiiii

⋅+⋅+⋅

∑

hps

FC: Fixed cost per unit of initial collection point

n: Number of initial collection points

VC: Variable cost of initial collection point

LC: Logistic cost per unit of returned product

OC: Other costs

After roulette wheel mate selection, one point

crossover process (crossover rate 0.8) takes place.

The crossover process selects a string randomly in

the intermediate solution space and modifies its gens

value randomly. This is to ensure that all possible

solutions are explored and also to avoid the solution

space from converging to a local maxima or minima.

Example of how the mutation operator works:

parents offspring

1 1 --- 1 0 0 0 1 1 1 --- 1 10 0 0

0 0 --- 1 10 0 0 0 0 --- 1 0 0 0 1

Exchange site

After the crossover operation, 2 new chromosomes

will be formed and this will ensure diversity in the

solution phase. Next step is mutation operation with

mutation rate as 0.01. Unlike crossover operation,

mutation involves only 1 gene. The operator

randomly selects gens and assigns random value to it

as shown below:

Before After

1 1

1 0 1 0 1 0 1 0 1 0 1 1 1 0

The random gene alteration process also works to

ensure that diversity in the solution set is maintained

ICEIS 2009 - International Conference on Enterprise Information Systems

296

and that the solution do not converge pre-maturely

to a local maxima or minima.

Finally is the replacement process. After the

mutation has taken place, the chromosomes are then

evaluated again against the objective function.

Chromosomes with higher scores will then be

chosen into the next round of mutation process while

the chromosomes with lower scores are discarded.

Through these processes, the solution space will

converge to the most optimum solution desired.

Design Framework. As shown in Figure 2 and 3,

using the genetic algorithm generator program, we

could determine the location of initial collection

point to achieve the best possible population

coverage.

Figure 2: Genetic Algoritm Generator.

Figure 3: Determination of initial collection point location.

However, this method of optimization is based on an

ideal case scenario. The program does not take into

account factors such as accessibility and

convenience to the end users. Therefore, the model

above is merely an indication of the approximate

location. Even so, we could still employ the

knowledge acquired above to be applied when

making a final decision on the placement of the

collection center.

Furthermore, a collection schedule will have to

be planned for regular pick-up at the initial

collection center back to the centralized collection

center. Due to the limited space available at the

initial collection centers, pick-up schedule has to be

regular and frequent. Planning of the pick-up

schedule is not covered in this paper.

4.2.3 Centralized Collection Center

The centralized collection center is a place where all

the return products will be sent to from the initial

collection center. This is where the returned

products will be sorted, stored and distributed to

undergo their respective process. Hence, the

centralized collection center is where the workflow

begins and is where we shall employ RFID

technology to optimize the efficiency.

Radio Frequency Identification (RFID) for

Reverse Logistics. Radio Frequency Identification

or RFID has moved from upcoming technology to

mainstream application in recent years. RFID is

currently been used to replace the tradition bar-code

system as it can read data without the need of being

in line-of-sight and the RFID tags can store much

more data than bar codes. Among the earliest

adopters of RFID includes Wal-Mart, Tesco and the

US Department of Defence (Want., 2007). However,

most of the usage of RFID at them moment is

mainly focused on the forwards logistics. In this

paper, we will integrate RFID technology into the

reverse logistics system to overcome problems faced

in the system.

RFID Working Principles. The very fundamental

working of RFID technology is based on

electromagnetic wave (EM). There are 2 different

types of RFID designs, mainly: Near field RFID and

Far field RFID. Near field RFID operates based on

Faraday’s principle of magnetic induction. This

system uses EM waves to power the RFID tag and

the tag transmits data back to the reader through load

modulation. Range of operation for this system

varies inversely with the frequency of operation

while energy from induction reduces with distance.

Far field RFID on the other hand captures the EM

wave propagated from the reader and reflects back

the EM wave through the embedded antenna on the

tag. This method of operation is known as back

scattering. The reader will then receive the reflected

wave pattern and interprets the data. Operational

range for this system is limited by the amount of

energy received by the tag from the reader.

DESIGN A REVERSE LOGISTICS INFORMATION SYSTEM WITH RFID

297

Types of RFID Tags. There are 2 types of RFID

tags: active and passive. Active RFID tags require

power source to operate. This power source can be

either from a power infrastructure or an integrated

battery. Since the tags require external power source,

they have limited lifespan. On the other hand,

passive tags receive power from the RFID reader.

Hence they are only activated upon being read by

the reader and have unlimited lifespan. Furthermore,

RFID tags shown in Figure 4 come in different

shape and sizes for different usage purposes. There

are also different types of tags for use on different

material surfaces.

Figure 4:Different shape and sizes of RFID Tags (Want.,

2007).

Advantages of RFID for Reverse Logistics. RFID

technology provides numerous advantages for the

proposed model in reverse logistics processes.

Among the advantages are:

• No line-of-sight needed. This enables data to

be read from the tag in returned products at any

orientation and hidden away from sight.

• Long-range reading is also possible with RFID.

This enables the reader to be at a distance away

from the tag while still being able to read the

tag. This is an added advantage for logistics

purposes.

• Multiple tag reading is also supported by

RFID. As compared to bar-code system, which

can only read 1 bar code at any one time, RFID

can read multiple tags at the same time. This

will enable the returned flow of products be

read at once when passing through the RFID

readers.

• RFID technology also allows for real time

tracking. When the returned product is in range

of the RFID reader, it will be read immediately

without human intervention. The data read by

the reader can subsequently be sent to the

respective actors involved in the reverse

logistics process.

RFID Constraints in Reverse Logistics. Even

though RFID provides a huge amount of advantages

over traditional system, it is not without its own

disadvantages. Chief among them are:

• Orientation of tags relative to the reader.

Although RFID tags do not require line-of-

sight for reading, however under certain

orientations, RFID tags can be hard to read.

This will pose a problem as the returned

products would be oriented randomly and this

could prevent the reader from reading the tags

effectively.

• Signal blockage. If the returned product is

made of metal casing, the RFID tag could be

enclosed in the casing causing EM wave to be

unable to penetrate the metal case and hence

will not be able to be read. This constraint is

particularly pronounced when tagging

components that are enclosed in a casing such

as computers, television sets and others.

• Cost. At the time being, the cost of RFID

hardware is still high. Moreover, reverse

logistics is usually seen as a cost center for

most companies and are not revenue

generating. Hence, companies are reluctant to

invest huge amount of money in this

technology. However, this cost is dropping as

the system gets more commercialized and more

companies adopt the technology.

• Privacy. Privacy is also a concern for end

users, as they fear that companies can still track

the products after consumers have purchased it.

However, certain measures have been taken to

address this issue.

In this paper, we have worked within the limitations

imposed by RFID technology to implement RFID to

design a better reverse logistics management system.

At the centralized collection center, RFID system

can be set up at the point of entry to scan each and

every collected product that has been tagged. An

important assumption made in this research is that

tagging is done at the forward logistics process and

is carried down to the reverse logistics process. The

RFID reader can read every tag passing through the

entry point and the system will relay this

information to the control centre.

With this data acquired through the RFID reader,

we are able to tell the quantity and type of product

returned in real time and without much human

intervention. This solves part of the unknowns in the

reverse logistics process. With the data acquired,

different products can be sorted according to their

respective category and this reduces time wastage in

between handling.

ICEIS 2009 - International Conference on Enterprise Information Systems

298

4.2.4 Producers

Fianally, the loop ends at the producer of products

who receive the returned flow of material and

convert it back to usable products to be sold back to

the end users. This cycle will repeat and make up the

reverse logistics process.

4.3 Case Example

As a summary of the whole process, a company (eg.

Apple Computers) can employ RFID technology to

tag all of their products during manufacturing. This

RFID tag will be used during the forward logistics to

replace bar-code technology. The tag will remain

with the product until the consumer decide to

dispose off of the product. Hence, the consumer will

return the product to the initial collection center

situated at various location throughout the area. A

routine collection schedule will ensure that returned

products are collected regularly from the initial

collection center back to the centralized collection

center. Once back in the center, the RFID reader

situated at the entrance will read the tags on the

returned product and disseminate the data to the

control center. This data will then be used to activate

the various work flow associated with each and

every returned product. Finally, the returned product

can be recycled and sent back to the original

producer to be made into new products and

distributed again.

5 CONCLUSIONS

In this paper, we formulated a model that integrates

Radio Frequency Identification (RFID) into reverse

logistics management in order to eliminate part of

the unknowns in the process. By doing this, we are

able to better plan and optimize the reverse logistics

network. Furthermore, we also utilized genetic

algorithm optimization technique to determine the

optimum location of initial collection points. This

will enable us to further reduce logistical cost

involved in reverse logistics. However, further study

about model evaluation with real-time data should

be done and comparing GA with other optimization

techniques can be carried out to show the

effectiveness of GA. We believe this paper could be

a reference point for further research to be

performed upon.

ACKNOWLEDGEMENTS

We wish to acknowledge the funding support for

this project from Nanyang Technological University

under the Undergraduate Research Experience on

Campus (URECA) programme.

REFERENCES

ReturnBuy. The new dynamics of returns: the profit,

customer and business intelligence opportunities in

returns. Ashburn, Virginia; White Paper; 2000,

ReturnBuy.com.

M. Fleischmann, J.M. Bloemhof-Ruwaard, R. Dekker, E.

V. D. Laan, Jo A.E.E. van Nunen, L. H. Van

Wassenhove, “Quantitative models for reverse

logistics: A review”, European Journal of Operational

Research 103, 1997, pp. 1-17.

R. Want, “An introduction to RFID technology”,

Pervasive Computing, January-March 2006, pp 25-33.

V. D. Hunt, A.Puglia, M. Puglia, RFID: A Guide to Radio

Frequency Identification. Wiley-Interscience, 2007.

H. Min, H. J. Ko, C. S. Ko, “A genetic algorithm apporach

to developing the multi-echelon reverse logistics

network for product returns”, Omega: The

International Journal of Management Science, vol 34,

2003, pp. 56-69.

DESIGN A REVERSE LOGISTICS INFORMATION SYSTEM WITH RFID

299