AN ENHANCED SERVICE PROVIDER COMMUNICATION

INTERFACE WITH CLIENT PRIORITIZATION

Case Study on Fast-food Chain Restaurants

Slobodan Lukovic

ALaRI, University of Lugano, Lugano, Switzerland

Nikola Puzovic

Department of Information Engineering, University of Siena, Siena, Italy

Milos Stanisavljevic

Microelectronic Systems Laboratory, EPFL, Lausanne, Switzerland

Keywords:

e-Commerce, client prioritization, mobile devices, workload scheduling.

Abstract:

With the increased dynamics of modern life, the efﬁciency and reliability of everyday services is emerging to

be a fundamental concern. On the other hand, modern telecommunication technologies, like wireless Internet

access, are penetrating all segments of our life. However, many every day activities and services still do not

fully exploit new technologies. We propose an approach that enables increased deployment of E-commerce

concepts in the ﬁelds where their usage was either small or negligible. Moreover, in the scope of the same

concept, we introduce prioritization of clients in services where it was not commonly present to date.

A solution for enhanced communication interface between service provider and customers is developed. As

a case study, the system is designed and optimized for an implementation in a fast-food chain. The proposed

solution is aiming at increasing of quality of service for customers, and at the same time increasing the op-

erational efﬁciency of the provider. The main idea behind this approach is to enable customers to use their

mobile devices, such as cell phones or PDAs, for browsing offered services or goods, viewing current service

conditions and placing orders. We will detail theoretical concepts underneath and describe the implementation

on both server and client side.

1 INTRODUCTION

Millions of people daily face different kinds of prob-

lems in common situations. These issues range

from unpredictable trafﬁc and parking problems to

big waiting times in restaurants due to inefﬁcient

order/payment service. The rapidly growing urban

population additionally increases pressure on service

providers. We look for a response for these problems

by incorporating e-Commerce concepts into service

provider - client chain.

The basic idea of service order/payment process

automation relies on the rapid increase of number of

portable devices that are able to access Internet, and

on the growth of the popularity of e-payment meth-

ods. It is estimated that inthe next two years all manu-

factured cell-phoneswill be equipped with WiFi mod-

ules which will boost the availability of Internet ser-

vices. Other broadband wireless services like UMTS

and WiMax are becoming more widely available and

affordable and the number of mobile users that will

use these services will increase. Moreover, the num-

ber of online payments is constantly increasing and

this way of payments is expected to be widely ac-

cepted in very near future. Putting all together, inte-

gration of all aforementioned services is logical con-

sequence of technology development and of the evo-

lution of users’ habits.

In this paper we detail, based on case study on

fast food chain, a possible solution for incorpora-

tion and automation of the service scanning, or-

dering, payment and execution that increases efﬁ-

ciency, cutting the costs and bringing many other

beneﬁts for all sides involved in the process. Our

aims are the improving of existing Quality of Ser-

vices (QoS) in fast food restaurants, and at the same

197

Lukovic S., Puzovic N. and Stanisavljevic M. (2008).

AN ENHANCED SERVICE PROVIDER COMMUNICATION INTERFACE WITH CLIENT PRIORITIZATION - Case Study on Fast-food Chain Restaurants.

In Proceedings of the International Conference on e-Business, pages 197-202

DOI: 10.5220/0001912401970202

 SciTePress

time providing greater ﬂexibility for both the ser-

vice provider and consumers. We propose a fully

integrated system that incorporates many different

technological aspects, ranging from Internet brows-

ing and e-payments to service performance evalua-

tion and workﬂow scheduling. The ﬁnal result is a

novel communication interface that brings many new

features and beneﬁts to the client, like better service

overview, waiting time prediction (service availabil-

ity overview),prioritizing of clients, multilingual sup-

port, etc. On the other side, it cuts costs for service

providers in terms of staff reducing, better insight in

demand and market overview coupled with market

proﬁling and targeted marketing. For now no experi-

mental results are provided.

The paper is organized as follows: In Section 2

we summarize current state of technology and differ-

ent types of its usage that are in scope of interest of

this work, in Section 3 we present an overview of the

overall architecture of the system. Algorithms and

techniques used in realization of proposed concepts

are detailed in Section 4. Finally, Section 5 presents

conclusions and future work.

2 STATE OF THE ART

In parallel to development of communication tech-

nologies many different service providers have been

adopting them to facilitate the interaction with clients,

to provide new services or increase operational efﬁ-

ciency. On the other side mobile technologies such

as WiFi, UMTS or WiMax as well as web access

standards and protocols are more and more oriented

toward better support for increased need for mo-

bile Internet availability. In that sense many mo-

bile web standards such as Wireless Markup Lan-

guage (WML), Extended Modeling Language (XML)

or XHTML etc. have been developed. Moreover, the

development of AJAX (AJAX, 2008) has enabled ex-

change of small amounts of data between client and

server, hence increasing the interactivity, speed and

usability without the need to reload the entire contents

of web page. This is especially valuable when de-

vices have scarce computational and communication

capabilities. By having more and more mobile de-

vices online many new concepts got enabled. This es-

pecially concerns social networking (Eagle and Pent-

land, 2005), mobile commerce (Varshney et al., 2000)

or intelligent wireless web (Alesso and Smith, 2001).

The most recent world-wide trend regarding wire-

less Internet access is deploying of WiFi Internet ac-

cess at variety of locations such as airports, hotels,

restaurants and so on (Friedman and Parkes, 2003);

in some cases even free of charge (Smithers, 2007).

At the same time online payment methods are getting

widespread across wide range of activities (Weiner,

2000),(Ghosh and Li, 2007). Coffees and restaurants

have made steps towards exploitation of WiFi at local

service points (Friedman and Parkes, 2003) and some

fast-food chains started using touch-screens deployed

at tables (eTable) for offer browsing and order placing

(VanLeeuwen, 2005).

We propose a novel solution for prioritization of

clients using and enhancing different experiences and

implementing them in ﬁelds where until now these

technologies have not been commonly used. In order

to provide better quality of service we model priori-

tized orders execution and service provider capacity

with a well know operating systems task scheduling.

The theoretical concepts adopted for purposes of our

work will be discussed in detail in Section 4.

3 SYSTEM OVERVIEW AND

IMPLEMENTATION

CONCERNS

We envision a system that integrates WLAN access,

priority order scheduling based on demand prediction

and delivery automation that provides user-friendly

interface to the client. For the purpose of the client-

server communication we propose WiFi wireless In-

ternet access technology since it is widely used and

already deployed at many fast food points around the

world. WiFI access represent an optimal compro-

mise between simplicity and efﬁciency. The cover-

age, throughput and level of security are considerably

greater than in the case of Bluetooth. On the other

side it is very easy to deploy and cheap to use (from

clients’ point of view it is free) in comparison to other

broadband services as UMTS and WiMax. The range

of the WiFi AP also makes it perfectly suitable for use

in a local service point (a restaurant in this case). Nev-

ertheless, the system is conceived in such way that it

can be easily ported to other communication medi-

ums. By using secure communication and already es-

tablished methods of electronic payment, we will also

provide high level of security that is necessary.

Figure 1 shows the overall organization of the sys-

tem. The main components are servers in service

points, central data warehouse and a connection to

e-payment servers. The local servers are responsi-

ble for processing clients’ orders and updating cen-

tral data warehouse. Payments are performed using

secure connection to e-payment servers or by using

in-house e-payment system. In the following subsec-

ICE-B 2008 - International Conference on e-Business

198

Internet

e-payment

servers

Central data

warehouse

Local server

Service Point 1

Service Point n

Figure 1: Proposed implementation of the system.

tions and we will also discuss the system from the

participants’ point of view (both clients’ and service

providers’), we will discuss the technology involved

together with different implementation aspects.

3.1 Client’s View of the System

Client devices are PDAs or smart phones that are ca-

pable of running a web browser. User interface is im-

plemented in HTML, and it is accessible from each

web browser, so there is no need to install a special

purpose application. Figure 2 shows the UML se-

quence diagram of the users’ interaction with the sys-

tem. At the beginning, the client chooses appropri-

ate WiFi Access Point (AP) that will consent him to

access the services provided in the restaurant. Upon

the entry to the network, clients are redirected to the

start page from where they can browse the available

services and current conditions of the system. There

is a possibility to supply the client with additional

information that traditional ordering methods cannot

support (i.e. expected waiting time, multilingual in-

terface etc.). The calculated expected waiting time

coupled with delivery automation brings a possibility

of introducing prioritization in the fast food service.

For more details see Section 4.

From the start page clients will continue to the

menu with products that are available in the restau-

rant. Trough simple web forms clients can make a

choice, and communicate the selection to the server

in the service point when the order is completed. The

system sends back the information about the order

and order number. The payment can be done through

SSL secured connection using credit-card payment,

pay-pal and other methods of e-payment. In this case,

a request for payment is communicated to the pay-

ment gateway (paypal server or bank server in case

of credit card payment) that processes it and gives the

conﬁrmation. Another option is to use vendor pro-

vided vouchers that can be issued in the form of ﬁ-

ServerUser e-payment server

Enter Network

Send initial web page

Send order

Send e-payment request

Payment processed

Order processed

Browse and

select services

Prepare data

Process request

Redirect

Send e-payment credentials

Process

payment

Confirm

order

Figure 2: Interaction between the client and the system.

delity card, and can be used exclusively for payments

inside the system, and they function in form of deposit

or credit.

Once the payment is completed and the delivery is

ready, it is placed in a delivery slot and the client re-

ceives a code associated with the order. The client

types the code using keyboard on the delivery slot

that is labeled with appropriate order number or with

clients name. These slots can be implemented as

parts of rotating table, where each slot has a protec-

tive cover. Once the correct code is inserted, the ser-

vice is considered to be completed. The time between

order placement and code insertion is considered as

’service time’. This information is taken and used as

correction factor for statistical processing and for cal-

culation of the expected time of servicing.

3.2 Service Provider’s View of the

System

The system in a service point processes e-Commerce

orders in parallel with traditional ones (Figure 3). Re-

quests from both sides go to the same server and their

execution is scheduled in order of submission. The lo-

cal server that processes orders is responsible for han-

dling client requests and for scheduling the delivery of

orders. Scheduling and delivery are performed taking

into account also the priority of the clients. The al-

gorithm for calculating waiting times is applied each

time a new client enters the restaurant and this infor-

mation is communicated to the client.

The technology used in the service point is rather

simple and cheap, and it involves a server computer,

wireless AP and Internet connection that will be used

for communication with the central server. The algo-

rithms that are used are described in subsequent sec-

tions.

AN ENHANCED SERVICE PROVIDER COMMUNICATION INTERFACE WITH CLIENT PRIORITIZATION - Case

Study on Fast-food Chain Restaurants

199

Figure 3: Overview of the service points’ organization.

4 SERVICE DEMAND AND

SERVICE TIME PREDICTION

MODEL

In this section we present mathematical concepts pro-

posed for calculation of expected average service time

considering prioritized clients. The main contribution

reﬂects in the exploration of scheduling mechanisms

in order to meet the deadlines - guarantee service. We

brieﬂy expose possible service classiﬁcation concepts

and compare them. The chosen concept is described

in detail.

4.1 Deﬁnition of the Server and Client

We consider the server - service execution unit that

takes the order, executes it and performs the deliv-

ery of the requested service or product. The server

is characterized as multiple parallel process execution

unit with a given total capacity - service power. The

clients could be clasiﬁed in three groups with two lev-

els of priority:

1. Traditional clients served at the service line in

a traditional way (worker at the counter table is

serving one client after another from the FIFO -

ﬁrst in, ﬁrst out, queue), taking low priority (pri-

ority equal 0).

2. Regular e-Commerce clients, performing orders

through wireless system, taking low priority.

3. Prioritized e-Commerce clients, performing prior-

ity service orders through wireless system, taking

high priority (priority equal 1).

Initially a certain server capacity is given to the prior-

itized clients. This capacity is dynamically adjusted

during service execution in order to guarantee pro-

jected waiting time for prioritized clients and there-

fore quality of service (QoS). We have adopted fol-

lowing naming conventionto describe different issues

related to times spent in different phases of the order

processing process:

• Queuing time - time spent in the line/queue while

waiting for the order placing.

• Service execution time - time required to execute

the requested service by the server.

• Service waiting time - the time that passes from

the moment of performing the order to the mo-

ment in which the order is completed.

• Total service time - the time from the moment

the client has entered service point to the moment

when the required services are obtained.

4.2 Classiﬁcation of Services by

Processing Power Requirements

Almost all orders differ from each other in type of

services requested, its quantity and quality which re-

sults in different service processing power require-

ments. Any order can be further decomposed in set of

’atomic services’. We assume ’atomic service’ to be

the simplest possible single order (i.e. an ice cream

or coffee). In that sense every order can be seen as

a composition of various ’atomic orders’. The most

precise way of modeling the orders would be repre-

senting them in form of number of atomic services

requests. Unfortunately this method would introduce

huge computational overhead and the entire system

would be very complex to implement. For the sake

of the simplicity and better efﬁciency but without los-

ing generality we classify orders according to number

of atomic services requested into ﬁve classes ranging

from small to huge ones. Each of those order types

are assigned certain evaluated average service time.

This evaluation is constantly updated by newly ob-

tained data from the system.

4.3 Service Time Prediction Model and

Scheduling for a Guaranteed

Execution

Here we describe empirical model for run-time calcu-

lation of service expected time. This time is provided

to clients as additional information and this property

of the system is one more original feature that en-

hances existing services. The service time is modeled

as a statistical variable with two components:

1. Non-stochastic component measuring the neces-

sary service time for non-prioritized and priori-

tized services with a given scheduling scheme.

ICE-B 2008 - International Conference on e-Business

200

2. Stochastic component that incorporates the addi-

tional service time related to the average num-

ber of prioritized clients that are arriving during

the time slot deﬁned with non-stochastic compo-

nent. The additional service time is added into

the scheduling scheme as a prioritized process and

a total service time is acquired after the probe

scheduling execution. This componentis acquired

by processing empirical data related to number of

prioritized clients and their orders that are dynam-

ically updated every day. During the phase of ini-

tial system deployment these data are unavailable

and therefore a worst case estimation is taken in-

stead of statistical component.

The order execution inside the server is scheduled ac-

cording to non-preemptive task execution scheduling.

We consider this scheme to be the most suitable in

case of fast food restaurants as preemptive service

would cause disorder in execution of currently exe-

cuted services and also would require different work-

ing concept inside the server (i.e. specialization to

atomic tasks execution and work division that could

support this would result in inefﬁcient use of working

power).

In case that prioritized tasks are not schedulable

within dedicated server capacity for prioritized tasks,

the amount of dedicated server capacity is increased

to the minimal value that guarantees schedulability of

prioritized tasks. This has a consequence of increase

in waiting time for non-prioritized clients. However,

QoS for prioritized clients is guaranteed.

Any practical scheduling algorithm in multipro-

cessing server systems presents a trade-off between

performance and computational complexity. How-

ever, in our case scheduling computation time is not

an issue (because it can be in the range of seconds)

and we can explore the more complex scheduling

schemes. Scheduling could be regarded as soft real-

time or even non real-time problem. The Earliest

Deadline First (EDF) algorithm is the most widely

studied scheduling algorithm for real-time systems

(Balarin et al., 1998).

EDF is more efﬁcient than many other schedul-

ing algorithms, including the static Rate-Monotonic

scheduling algorithm. However, when the process-

ing server is overloaded (i.e., the combined require-

ments of pending tasks exceed the capabilities of the

system) EDF performs poorly. Researchers have pro-

posed several adaptive techniques for handling heav-

ily loaded situations, but they require the detection of

the overload condition. Least Laxity First (LLF) al-

gorithm (Ramamritham and Stankovic, 1994) is non-

preemptive and selects the task that has the lowest

laxity (the maximum time that a task can wait before

Subtraction

Laxity

Priority

Server

Capacity

Time

Interface Engine

Total

Service

Time

Execution

Time

Run

Time

Priority

Figure 4: Inference model

being executed; laxity = total service time - service

execution time) among all the ready ones whenever

a processing server becomes idle, and executes it to

completion.

Lee et al. (Lee et al., 1994) presents a fuzzy

scheduling algorithm. Their proposed algorithm uses

task laxity and task criticality as system parameters

and doesn’t consider fairness. Their simulation model

contains small number of tasks on a uni-processing

unit system and they did not consider system over-

loads.

Chen et al. (Chen et al., 2005) proposed a schedul-

ing model and a related algorithm that is suitable

for both uni-processing and multiprocessing servers.

They provide a method to detect work overloading

and try to balance load with task dispatching. We pro-

pose to use model presented in (Hamzeh et al., 2007)

using a fuzzy interface engine. The model we pro-

pose has a slight modiﬁcation considering that there

are only two levels of priority deﬁned as ”high” and

”low”. As shown in Figure 4, the major factors con-

sidered in used approach to determine the scheduling

are task priority, total service time, service execution

time, and used server capacity time. The notion of

laxity is used in the proposed approach to facilitate

the computation.

In proposed algorithm as shown in Figure 5, a

newly arrived task will be added to the input queue.

This queue contains the remaining tasks from last cy-

cle that has not yet been assigned.

Fuzzy scheduler processes each task separately,

computes its run-time priority and sends it to task dis-

patcher’s priority queue. In a multiprocessing system,

this queue offers tasks to dispatcher by their run-time

priority order (as shown in Figure 6). Dispatcher of-

fers a new task whenever one of the processing units

of the system ﬁnishes its current task.

AN ENHANCED SERVICE PROVIDER COMMUNICATION INTERFACE WITH CLIENT PRIORITIZATION - Case

Study on Fast-food Chain Restaurants

201

1. For each task in input queue

(a) Feeds task’s run-time priority using fuzzy infer-

ence engine

2. While a server has a free processing unit

(a) assign the task with highest run-time priority to

the processing unit

3. Loop forever

(a) If a processing unit event occurs endenumerate

i. Go to 2

(b) If scheduling event occurs

i. Update tasks parameters

ii. Go to 1

Figure 5: Proposed algorithm

Task Queue

Task Scheduler

Task Dispatcher

Process Unit

Priority Queue

Figure 6: System view of real-time fuzzy scheduler

5 CONCLUSIONS AND FUTURE

WORK

In this work we have proposed a solution for deploy-

ment of e-Commerce concept in the fast food restau-

rant chain that brings more convenience for both ser-

vice provider and clients. The novelty of the work

lays in possible implementation of client prioritiza-

tion based on a well-known computer science concept

of operating system task scheduling, that will perform

the best under the assumption that we have enough

information considering stochastic component of ser-

vice time. For such a credible statistics we need to

have the insight of a system in a long run. The pro-

posed system brings numerousbeneﬁts to both parties

involved in the service process.

The system will boost the efﬁciency of the ser-

vice by eliminating ordering waiting time and will cut

costs by decreasing staff needed for order acceptance

and delivery performing. It will also increase the vis-

ibility (if coupled with Internet - browsing and posi-

tioning) of the services, and enables better demand in-

sight that brings more ﬂexibility. Moreover, it will en-

able market proﬁling and targeted proﬁling, and gives

the possibility to offer more e-services according to

clients’ need.

There are many beneﬁts for the clients also. They

will achieve precise insight in offer and service con-

dition, the queuing time will be eliminated and they

will have the possibility to get delivery in short time

as prioritized users. They will also have more pay-

ment options, and multilingual interface.

Our future work will focus on collecting and

thoroughly analyzing statistical data. In the early

phase of system deployment, realistic assumptions for

worst case scenario needs to be made. Also, tun-

ing of scheduling algorithm needs to be performed

with a detailed testing with realistic data versus other

scheduling schemes.

REFERENCES

AJAX, A. (2008). http://www.asp.net/ajax/.

Alesso, P. and Smith, C. (2001). The intelligent wireless

Web. Addison-Wesley Prof., London, 2nd edition.

Balarin, F., Lavagno, L., Murthy, P., and Sangiovanni-

vincentelli, A. (1998). Scheduling for embedded real-

time systems. IEEE Design and Test of Computers,

15(1):71–82.

Chen, G., Chen, G., Ozturk, O., and Kandemir, M. (2005).

An adaptive locality-conscious process scheduler for

embedded systems. In Proc. 11th IEEE Real-Time and

Embedded Technology and Applications Symposium.

Eagle, N. and Pentland, A. (2005). Social serendipity: Mo-

bilizing social software. IEEE Pervasive Computing,

4(2):28–34.

Friedman, E. and Parkes, D. (2003). Pricing wiﬁ at star-

bucks – issues in online mechanism design. In Proc.

Fourth ACM Conf. on Elec. Commerce (EC’03).

Ghosh, S. and Li, X. (2007). Critical factors affecting

the acceptance of e-payment services in emerging

economies. In Proc. Decision Sciences Institute Con-

ference.

Hamzeh, M., Fakhraie, S. M., , and Lucas, C. (2007).

Soft real-time fuzzy task scheduling for multiproces-

sor systems. International Journal of Intelligent Tech-

nology, 2(4):211–216.

Lee, J., Tiao, A., and Yen, J. (1994). A fuzzy rule-based

approach to real-time scheduling. In Proc. 3rd IEEE

Conf. Fuzzy Systems.

Ramamritham, K. and Stankovic, J. (1994). Scheduling al-

gorithms and operating systems support for real-time

systems.

Smithers, R. (2007). Mcdonald’s to offer free wi-ﬁ in

restaurants. Guardian.

VanLeeuwen, S. (2005). Taking food service into 21th cen-

tury. In eTable Inc.

Varshney, U., Vetter, R. J., and Kalakota, R. (2000). Mobile

commerce: A new frontier. Computer, 33(10):32–38.

Weiner, S. E. (2000). Electronic payments in the u.s. econ-

omy: An overview. Computer.

ICE-B 2008 - International Conference on e-Business

202